Classifications

• • G—PHYSICS
  • G02—OPTICS
  • G02B—OPTICAL ELEMENTS, SYSTEMS OR APPARATUS
  • G02B7/00—Mountings, adjusting means, or light-tight connections, for optical elements
  • G02B7/28—Systems for automatic generation of focusing signals
  • G02B7/36—Systems for automatic generation of focusing signals using image sharpness techniques, e.g. image processing techniques for generating autofocus signals
• • G—PHYSICS
  • G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
  • G03B—APPARATUS OR ARRANGEMENTS FOR TAKING PHOTOGRAPHS OR FOR PROJECTING OR VIEWING THEM; APPARATUS OR ARRANGEMENTS EMPLOYING ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ACCESSORIES THEREFOR
  • G03B13/00—Viewfinders; Focusing aids for cameras; Means for focusing for cameras; Autofocus systems for cameras
  • G03B13/32—Means for focusing
  • G03B13/34—Power focusing
  • G03B13/36—Autofocus systems
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
  • H04N23/60—Control of cameras or camera modules
  • H04N23/67—Focus control based on electronic image sensor signals
  • H04N23/673—Focus control based on electronic image sensor signals based on contrast or high frequency components of image signals, e.g. hill climbing method

Definitions

• the subject matter disclosed generally relates to auto- focusing electronically captured images.
• Photographic equipment such as digital cameras and digital camcorders may contain electronic image sensors that capture light for processing into still or video images, respectively.
• Electronic image sensors typically contain millions of light capturing elements such as photodiodes.
• the process of auto-focusing includes the steps of capturing an image, processing the image to determine whether it is in focus, and if not, generating a feedback signal that is used to vary a position of a focus lens ("focus position") .
• focus position a position of a focus lens
• the other technique looks at a phase difference between a pair of images.
• the contrast method the
• the phase difference method includes splitting an incoming image into two images that are captured by separate image sensors. The two images are compared to determine a phase difference. The focus position is adjusted until the two images match.
• the phase difference method requires additional parts such as a beam splitter and an extra image sensor.
• the phase difference approach analyzes a relatively small band of fixed detection points. Having a small group of detection points is prone to error because noise may be superimposed onto one or more points. This technique is also ineffective if the detection points do not coincide with an image edge.
• the phase difference method splits the light the amount of light that impinges on a light sensor is cut in half or even more. This can be problematic in dim settings where the image light intensity is already low.
• An auto focus image system that includes a pixel array coupled to a focus signal generator.
• the pixel array captures an image that has a plurality of edges.
• the generator may determine to reduce a relative extent to which an edge contributes to the focus signal on basis of detecting that the edge does not have sufficient reflection symmetry, according to a predefined criterion, in a sequence of gradients of an image signal across the edge.
• FIG. 1 is a schematic of an embodiment of an auto-focus image pickup apparatus
• FIG. 2 is a schematic of an alternate embodiment of an auto-focus image pickup apparatus
• FIG. 3 is a block diagram of a focus signal generator
• FIG. 4 is an illustration of a horizontal Sobel
• FIG. 5 illustrates a calculation of edge width from a horizontal gradient
• FIG. 6A, 6B are illustrations of a calculation of an edge width of a vertical edge having a slant angle ⁇ ;
• FIG. 6C, 6D are illustrations of a calculation of an edge width of a horizontal edge having a slant angle ⁇ ;
• FIG. 7 is a flowchart of a process to calculate a slant angle ⁇ and correct an edge width for a vertical edge having a slant ;
• FIG. 8 is an illustration of a vertical concatenated edge
• FIG. 9A is an illustration of a group of closely-packed vertical bars
• FIG. 9B is a graph of an image signal across FIG. 9A;
• FIG. 9C is a graph of a horizontal Sobel gradient across FIG. 9A;
• FIG. 10 is a flowchart of a process to eliminate closely- packed edges having shallow depths of modulation;
• FIG. 11 is a histogram of edge widths illustrating a range of edge widths for calculating a fine focus signal
• FIG. 12 is an illustration of a scene
• FIG. 13 is a graph illustrating a variation of a narrow- edge count during a focus scan of the scene of FIG. 12;
• FIG. 14 is a graph illustrating a variation of a gross focus signal during a focus scan of the scene of FIG. 12;
• FIG. 15 is a graph illustrating a variation of a fine focus signal across a range of focus positions;
• FIG. 16 is an illustration of an apparatus displaying multiple objects in a scene and a selection mark over one of the objects;
• FIG. 17 is a block diagram of an alternate embodiment of a focus signal generator;
• FIG. 18 is a schematic of an alternate embodiment of an auto-focus image pickup apparatus
• FIG. 19 is a schematic of an embodiment of an auto-focus image pickup apparatus having a main pixel array and an auxiliary pixel array;
• FIG. 20 is a schematic of an alternate embodiment of an auto-focus image pickup apparatus having a main pixel array and an auxiliary pixel array
• FIG. 21 is a schematic of an alternate embodiment of an auto-focus image pickup apparatus having a main pixel array and an auxiliary pixel array
• FIG. 22 is an illustration of a variation of an edge width from a main pixel array and a variation of an edge width from an auxiliary pixel array at different focus positions;
• FIG. 23A illustrates a gradient of an image signal across two adjacent edges of opposite polarities (i.e. signs) where the edges do not interact;
• FIG. 23B illustrates a gradient of an image signal across two adjacent edges of opposite polarities (i.e. signs) where the edges interact;
• FIG. 24A shows the positive gradients from FIG. 23B and illustrates that the distance between a pair of interpolated gradients at a particular gradient level is divided into two unequal distances by the interpolated peak;
• FIG. 24B shows the positive gradients from FIG. 23B and illustrates that the area between two gradient levels and bordered on the left and right by the two sides of the
• FIG. 24C shows the positive gradients from FIG. 23B and illustrates a length of a segment of the gradient profile between two gradient levels, an area of a region vertically under the segment and the lower gradient level, and a width of a base of the region;
• FIG. 24D shows the positive gradients from FIG. 23B and illustrates a method for estimating the first derivative
• FIG. 24E shows the positive gradients from FIG. 23B and illustrates an alternative method for estimating the first derivative
• FIG. 24F shows the positive gradients from FIG. 23B and illustrates a method for estimating the second derivative
• FIG. 24G shows the positive gradients from FIG. 23B and illustrates a alternative method for estimating the second derivative
• FIG. 24H shows the positive gradients from FIG. 23B and illustrates a distance between midpoints at different gradient levels and between each midpoint and the interpolated peak;
• FIG. 241 shows the positive gradients from FIG. 23B and illustrates two gradients at a common distance from the interpolated peak
• FIG. 24J shows a symmetric gradient profile
• FIG. 25 illustrates a sequence of second derivatives of an image signal across an edge plotted against distance in multiples of a spacing between successive second derivatives, showing (a) a width W s between a pair of positive and negative peaks, (b) a width Wi between a pair of outermost interpolated second derivatives that have a given magnitude h lr (c) a width i3 ⁇ 4> between an inner pair of interpolated second derivatives that have the given magnitude h lr and (d) a distance Di from a zero-crossing (between the pair of positive and negative peaks) to an outermost interpolated second derivative that has the given magnitude hi;
• FIG. 26 illustrates a sequence of image data samples of the image signal plotted against distance in multiples of a spacing between successive samples, showing (a) a width W edge and a contrast C eC fg e between two samples at two ends of the edge, (b) a peak gradient value g peak between a pair of samples that has a steepest change of sample value, (c) an undivided portion of the edge that has contrast Ci and width W partlr and (d) an undivided portion of the edge that has contrast C ⁇ and width W part2 ;
• FIG. 27 illustrates a sequence of gradients across an edge plotted against distance in multiples of a spacing between successive gradients, and an area of a region under the plotted sequence of gradients
• FIG. 28 illustrates a sequence of gradients of an image signal across an edge plotted against distance in multiples of a spacing between successive gradients, a center of gravity (i.e. center of moment), and distances of the gradients from the center of gravity;
• FIG. 29 illustrates finding an interpolated peak's position by interpolation
• FIG. 30 shows an alternate embodiment of a focus signal generator .
• an auto focus image system that includes a pixel array coupled to a focus signal generator.
• the pixel array captures an image that has at least one edge with a width.
• the focus signal generator may generate a focus signal that is a function of the edge width and/or statistics of edge widths.
• An auto focus image system that includes a pixel array coupled to a focus signal generator.
• the pixel array captures an image that has at least one edge with a width.
• the generator generates a focus signal that is a function of the edge width and various statistics of edge width.
• the generator may eliminate an edge having an asymmetry of a gradient of an image signal.
• the generator may also eliminate an edge that fails a template for an associated peaking in the gradient.
• a processor receives the focus signal and/or the statistics of edge widths and adjusts a focus
• the edge width can be
• a histogram of edge widths may be used to determine whether a particular image is focused or
• a histogram with a large population of thin edge widths is indicative of a focused image.
• Figure 1 shows an embodiment of an auto-focus image capture system 102.
• the system 102 may be part of a digital still camera, but it is to be
• the system 102 may include a focus lens 104, a pixel array and circuits 108, an A/D converter 110, a processor 112, a display 114, a memory card 116 and a drive
• the motor/circuit 118 Light from a scene enters through the lens 104.
• the pixel array and circuits 108 generates an analog signal that is converted to a digital signal by the A/D Converter 110.
• the pixel array 108 may
• the digital signal may be sent to the processor 112 that performs various processes, e.g. color
• a color interpolation unit 148 may be implemented to perform color interpolation on the digital signal 130 to estimate the missing color signals on each pixel for the focus signal generator 120. Alternately, where the focus signal generator 120 and the processor 112 reside
• the focus signal generator 120 may input interpolated color images from the
• processor 112 on bus 146 as shown in Figure 2 or a single image signal derived from the original image signal generated from the A/D converter 110, for example a grayscale signal .
• the focus signal generator 120 receives a group of control signals 132 from the processor 112, in addition, and may output signals 134 to the processor 112.
• the output signals 134 may comprise one or more of the following: a focus signal 134, a narrow-edge count, and a set of numbers representing a statistics of edge width in the image.
• the processor 112 may generate a focus control signal 136 that is sent to the drive
• a focused image is ultimately provided to the display 114 and/or stored in the memory card 116.
• the algorithm(s) used to adjust a focus position may be performed by the processor 112.
• the pixel array and circuits 108, A/D Converter 110, focus signal generator 120, and processor 112 may all reside within a package. Alternately, the pixel array and circuits 108, A/D Converter 110, and focus signal generator 120 may reside within a package 142 as image sensor 150 shown in Figure 1, separate from the processor 112. Alternately, the focus signal generator 120 and processor 112 may together reside within a package 144 as a camera controller 160 shown in Figure 2, separate from the pixel array 108 and A/D Converter 110.
• the focus signal generator 120 (or any alternative embodiment, such as one shown in Figure 30) and the processor 112 may together reside on a semiconductor substrate, such as a silicon substrate.
• FIG 3 shows an embodiment of a focus signal generator 120 receiving image (s) from a image providing unit 202.
• the image providing unit 202 may be the color interpolator 148 in Figure 1 or the processor 212 in Figure 2.
• the focus signal generator 120 may comprise an edge detection & width measurement (EDWM) unit 206, a focus signal calculator 210, a length filter 212, and a width filter 209. It may further comprise a fine switch 220 controlled by input ⁇ fine' 222.
• the focus signal generator 120 may provide a narrow-edge count from the width filter 209 and a focus signal from the focus signal calculator 210, the focus signal being configurable between a fine focus signal and a gross focus signal, selectable by input ⁇ fine' 222. Alternately, both fine focus signal and gross focus signal may be calculated and output as part of output signals 134.
• the edge detection & width measurement unit 206 receives image (s) provided by the image providing unit 202.
• EDWM edge detection & width measurement
• control signals such as control signal ⁇ fine' 222, may be provided by the processor 112 in signals 132.
• the output signals 134 may be provided to the processor 112, which functions as a focus system controller that
• the focus position of the focus lens 104 controls the focus position of the focus lens 104 to bring images of objects into sharp focus on the pixel array 108 by analyzing the output signals 134 to detect a sharp object in the image.
• Various components of the focus signal generator 120 are described below.
• the EDWM unit 206 may transform the input image such that the three signals of the image, red (R) , green (G) and blue (B) are converted to a single image signal.
• RGB values can be used to calculate a luminance or chrominance value or a specific ratio of RGB values can be taken to form the single image signal.
• the single image signal may then be processed by a Gaussian filter or any lowpass filter to smooth out image data sample values among neighboring pixels to remove a noise.
• the focus signal generator 120, 120', 120" is not limited to grayscale signal. It may operate on any one image signal to detect one or more edges in the image signal. Or it may operate on any combination of the image signals, for example Y, R-G, or B-G. It may operate on each and every one of the R, G, B image signals separately, or any one or more combinations thereof, to detect edges. It may form statistics of edge widths for each of the R, G, B image signals, or any combination thereof. It may form a focus signal from statistics of edge widths from one or more image signals.
• the focus signal generator includes an edge detector to identify an edge in an image signal. The edge detector may use a first-order edge detection operator, such as Sobel operator, Prewitt operator, Roberts Cross operator, or Roberts operator.
• the edge detector may use a higher- order edge detection operator to identify the edge, for example a second order operator such as a Laplacian operator.
• the edge detector may use any one of the known edge detection operators or any improved operator that shares a common edge detection principle of any of the known operators.
• a gradient i.e. first derivative
• the edge detector uses a first-order edge detection operator
• a gradient i.e. first derivative
• There are various methods available to calculate the gradient including using any one of various first order edge detection operators such the Sobel operator, the Prewitt operator, the Roberts Cross operator, and the Roberts operator.
• the Roberts operator has two kernels which are single column or single row matrices: [-1 +1] and its transpose.
• Roberts Cross operator has two kernels which are 2-by-2 matrices: [+1, 0; 0, -1] and [0, +1; -1, 0], shown in the format of [ ⁇ first-row vector; second-row vector; third- row vector] like in Matlab.
• the Prewitt and the Sobel operator are basically have the same kernels, [-1, 0, +1] taking gradient in a direction of the row and its
• transpose taking gradient in a direction of the column, further multiplied by different lowpass filter kernels performing lowpass filterings perpendicular to the
• a second derivative (such as the Laplacian) of the image signal is computed.
• Each pixel may be tagged either a horizontal edge ( ⁇ ⁇ ' ) or a vertical edge ( ⁇ ' ) if either vertical or horizontal gradient magnitude exceeds a predetermined lower limit ("elimination threshold"), e.g. 5 for an 8- bit image, or no edge if neither is true. This lower limit eliminates spurious edges due to gentle shading or noise.
• a pixel may be tagged a vertical edge if its horizontal gradient magnitude exceeds its vertical gradient magnitude by a predetermined hysteresis amount or more, e.g. 2 for an 8-bit image, and vice versa. If both gradient magnitudes differ less than the hysteresis amount, the pixel gets a direction tag same as that of its nearest neighbor that has a direction tag already determined.
• a sequence of inspection of neighboring pixels may be the pixel above first, the pixel above left second, and the pixel on the left third, and the pixel above right last. Applying this hysteresis helps to ensure that adjacent pixels get similar tags if each of them has nearly identical horizontal and vertical gradient magnitudes.
• Figure 4 illustrates the result of tagging on a 6-by-6 array of horizontal and vertical gradients. In each cell, the horizontal gradient is in the upper-left, vertical gradient is on the right, and direction tag is at the bottom. Only pixels that have either horizontal or vertical gradient magnitude exceeding 5 qualify at this step as edge pixels are printed in bold and get direction tags.
• the image, gradients and tags may be scanned
• Each group of consecutive pixels in a same row, having a same horizontal gradient polarity and all tagged for vertical edge may be designated a vertical edge if no adjacent pixel on left or right of the group are likewise.
• each group of consecutive pixels in a same column having a same vertical gradient polarity and all tagged for horizontal edge may be designated a horizontal edge if no adjacent pixel above or below the group satisfies the same.
• horizontal and vertical edges may be identified.
• Each edge may be refined by removing pixels whose gradient magnitudes are less than a given fraction of the peak gradient magnitude within the edge.
• Figure 5 illustrates this step using a refinement threshold equal to one third of the edge's peak gradient magnitude, refining the edge width down to 3 from the original 9. This edge refinement distinguishes the dominant gradient component that sets the apparent edge width that
• Edge width may be calculated in any one of known methods.
• One method of calculating edge width is simply counting the number of pixels within an edge.
• FIG. 5 a first fractional pixel position (2.4) is found between a first outer pixel (pixel 3) of a refined edge and the adjacent outside pixel (pixel 2) by an interpolation from the refinement threshold 304.
• a second fractional pixel position (5.5) is found between a second outer pixel (pixel 5) and its adjacent outside pixel (pixel 6) .
• Another alternative edge width calculation method is to calculate a difference of the image signal across the edge (with or without edge refinement) and divide it by a peak gradient of the edge.
• edge width may be a distance between a pair of positive and negative peaks (or interpolated peak(s)) of the second order derivative of the image signal across the edge.
• edge-sharpness may be a distance between a pair of positive and negative peaks (or interpolated peak(s)) of the second order derivative of the image signal across the edge.
• edge-sharpness measure there are other alternatives than a width, which is merely one example of a edge-sharpness measure that is essentially independent of illumination of the scene.
• each edge may be assigned to one prescribed direction (e.g. vertical direction or horizontal
• direction e.g horizontal direction or vertical
• a boundary (shaded band) is shown to be inclined at a slant angle ⁇ with respect to the vertical dashed line, and a width a is shown to be measured in the perpendicular direction (i.e. horizontal direction) .
• a width b (as indicated in the drawing) measured in a direction perpendicular to the direction of the boundary (also direction of an edge that forms a part of the boundary) is more appropriate as the width of the boundary (and also of the edge) than width a .
• the edge widths measured in one or the other of those prescribed directions are to be corrected by reducing them down to be widths in directions
• the Edge Detection and Width Measurement Unit 206 performs such a correction on edge widths.
• the measured width a is the length of the hypotenuse of a right-angled triangle that has its base (marked with width b) straddling across the shaded boundary
• the corrected width b may then be obtained from a projection of the measured width a to the direction perpendicular to the edge direction. From elementary trigonometry, such a
• angle ⁇ or cos ( ⁇ ) itself, may be found by any method known in the art for finding a direction of an edge in an image, or by a more accurate method described in the flowchart shown in Figure 7.
• Each horizontal or vertical edge's edge width may be corrected for its slant from either the horizontal or vertical orientation (the prescribed directions),
• Figure 6A, 6B illustrate a correction calculation for an edge width measured in the horizontal direction for a boundary (and hence edges that form the boundary) that has a slant from the vertical line.
• Figure 6C, 6D illustrate a correction calculation for an edge width measured in the vertical direction for a boundary (and hence edges that form the boundary) that has a slant from the horizontal line.
• the correction may be made by multiplying the edge width measured in a prescribed direction, such as a vertical direction or a horizontal direction, by a factor of cos ⁇ , where ⁇ is an angle of slant from the prescribed direction.
• Figure 7 shows a flowchart of a process to correct edge widths for slant for edges inclined from a vertical line. (For horizontal edges, substitute 'row' for 'column' , and interchange 'vertical' with 'horizontal' in the flowchart.)
• a slant angle ⁇ is found. For each vertical edge, at step 502, locate the column position where the horizontal gradient magnitude peaks, and find the horizontal gradient x. At step 504, find where the vertical gradient magnitude peaks along the column position and within two pixels away, and find the vertical gradient y.
• the slant angle may be found by looking up a lookup table.
• step 508 scale down the edge width by multiplying with cos ( ⁇ ) , or with an approximation thereto as one skilled in the art usually does in practice.
• a first modification of the process shown in Figure 7 is to substitute for step 506 and part of step 508 by providing a lookup table that has entries for various combinations of input values of x and y. For each
• the lookup table returns an edge width correction factor.
• the edge width correction factor output by the lookup table may be an approximation to cos (tan -1 (y/x) ) to within 20%, preferably within 5%.
• the edge width is then multiplied with this correction factor to produce a slant-corrected edge width .
• a second modification is to calculate a quotient y/x between a vertical gradient y and a horizontal gradient x to produce a quotient q, then use q to input to a lookup table that has entries for various values of q. For each value of q, the lookup table returns an edge width
• the edge width correction factor may be an approximation to cos (tan (q) ) to within 20%, preferably within 5%.
• the values of x and y may be obtained in steps 502 to 506, but other methods may be employed instead.
• Adjacent edges may be prevented altogether from contributing to a focus signal, or have their
• Figure 9A, 9B, and 9C illustrate a problem that is being
• Figure 9A illustrates three vertical white bars separated by two narrow black spaces each 2 pixels wide.
• the middle white bar is a narrow bar 2 pixels wide.
• Figure 9B shows an image signal plotted horizontally across the image in Figure 9A for each of a sharp image and a blurred image.
• Figure 9C plots Sobel-x gradients of Figure 9B for the sharp image and blurred image.
• the first edge (pixels 2-5) for the blurred image is wider than that for the sharp image, and
• the two narrowest edges (pixels 9 & 10, and pixels 11 & 12) have widths of two in both images.
• the corresponding slopes at pixels 9 & 10, and pixels 11 & 12 each takes two pixels to complete a transition.
• the blurred image has a
• the minimum edge gap is in terms of a number of pixels, e.g. 1, or 2, or in between.
• edges may have been eliminated due to having a peak gradient less than the elimination threshold, two successive edges having an identical gradient polarity and spaced no more than two times the minimum edge gap plus a sharp_edge_width
• sharp_edge_width is a number assigned to designate an edge width of a sharp edge
• the Edge Detection and Width Measurement Unit 206 may execute the following algorithm for eliminating closely- packed narrower edges based on a screen threshold
• the screen threshold and screen flag to be used for the immediate next edge of an opposite polarity are determined according to the process of the flowchart shown in Figure 10.
• an edge may be eliminated unless one of the following conditions is true: (a) the screen flag is off for this edge, (b) a peak gradient magnitude of the edge is not smaller than the screen threshold for this edge.
• condition (c) the edge width is not less than sharp_edge_width + 1, where a number has been assigned for sharp_edge_width to designate an edge width of a sharp edge, and where the "+1" may be varied to set a range of edge widths above the sharp_edge_width within which edges may be eliminated if they fail (a) and (b) .
• sharp_edge_width may be 2.
• Figure 10 is a flowchart to determine a screen threshold and a screen flag for each edge. For vertical edges, assume scanning from left to right along a row, though this is not required. (For horizontal edges, assume scanning from top to bottom along a column, though this is not required.) A number is assigned for vertical edges. For vertical edges, assume scanning from left to right along a row, though this is not required. (For horizontal edges, assume scanning from top to bottom along a column, though this is not required.) A number is assigned for
• sharp_edge_width and may be 2 for the example shown in Figures 9A-9C.
• each edge is queried at step 720 as to whether its edge width is greater than or equal to one plus
• sharp_edge_width the value of one being the minimum edge gap value used for this illustration, but a different value may be used, such as between 0.5 and 2.0. If yes, the edge is a wider edge, and step 706 follows to set the screen threshold for the immediate next edge that has an opposite polarity to beta times a peak gradient magnitude of the edge, beta being from 0.3 to 0.7, preferably 0.55, then step 708 follows to turn on the screen flag for the next edge, then proceed to the next edge.
• step 730 follows to check whether the spacing from the prior edge of the same gradient polarity is greater than two times the minimum edge gap (or a different predetermined number) plus sharp_edge_width and the immediate prior edge of an opposite polarity, if any, is more than the minimum edge gap away. If yes, step 710 follows to turn off the screen flag for the next edge. If no, keep the screen flag and the screen threshold for the next edge and proceed to the next edge.
• Beta may be a predetermined fraction, or it may be a fraction calculated following a predetermined formula, such as a function of an edge width. In the latter case, beta may vary from one part of the image to another part.
• the image input by the focus signal generator 120 may have pixels laid out in a rectangular grid ("pixel grid") rotated at 45 degrees with respect to a rectangular frame of the image.
• pixel grid rectangular grid
• the X- and Y-directions of the edge detection operations and width measurement operations may be rotated likewise.
• Edge-sharpness measures In the above description, sharpness of image of an edge is represented by a width of the edge measured from a sequence of gradients across the edge with the
• edge-sharpness measure that is
• any edge- sharpness measure that has the above characteristic of being independent of or essentially independent of 20% scaling down of the image data in addition is a good alternative to the width measured from a gradient or interpolated gradient to another gradient or interpolated gradient of a same gradient value.
• the alternative edge-sharpness measure preferably has a unit that does not include a unit of energy.
• the unit of the edge-sharpness measure is determined on basis two points: (a) each sample of the image data on which the first-order edge-detection operator operates on has a unit of energy, (b) distance between samples has a unit of length. On basis of points (a) and (b) , a gradient value has a unit of a unit of energy divided by a unit of length. Likewise, contrast across the edge or across any undivided portion of the edge has a unit of energy.
• the contrast is not a good edge-sharpness measure, as the unit reveals that it is affected by illumination of the scene and reflectivity of the object. Neither is peak gradient of the edge, because the unit of the peak gradient has a unit of energy in it, indicating also that it is responsive to a change in illumination of the scene.
• peak gradient of the edge divided by a contrast of the edge is a good edge- sharpness measure, as it has a unit of the reciprocal of a unit of length.
• predetermine fraction of the peak gradient is a good edge-sharpness measure, as the count is simply a measure of distance quantized to the size of the spacing between contiguous gradients, hence having a unit of length.
• a gradient may be generated from a first-order edge detection operator used to detect the edge, or may be generated from a different first- derivative operator (i.e. gradient operator) .
• the Sobel operator or even a second-order edge detection operator, such as a Laplacian operator
• the Roberts operator whose kernels are simply [-1, +1] and its transpose, which is simply subtracting one sample of the image data from the next sample in the orientation of the gradient operator, with the resulting gradient located midway between the two samples.
• Edges may be detected with a higher-order edge detection operator than first-order independently of one or more derivative operators used in generating the edge-sharpness measure or any of the shape measures described in the next section.
• the edge-sharpness measure should have a unit of a power of a unit of length, for example a square of a unit of length, a reciprocal of a unit of length, the unit of length itself, or a square- root of a unit of length. Any such alternative edge-sharpness measure can replace the edge width in the focus signal generator 120.
• the correction factor as described above with reference to Figures 6A-6D and Figure 7 should be converted to adopt the same power.
• the edge-sharpness measure is peak gradient divided by a contrast, which gives it a unit of the reciprocal of a unit of length
• the appropriate correction factor for the edge-sharpness measure is the reciprocal of the correction factor described with reference to Figures 6A- 6D and Figure 7 above.
• the slant correction factor for the edge- sharpness measure should be a square of the width
• FIG. 27 illustrates a sequence of gradients across an edge plotted against distance in multiples of a spacing between successive gradients, and an area A3 of a shaded region under the plotted sequence of gradients.
• the region is defined between two gradient levels Li and L ⁇ , which may be defined with respect to an interpolated peak gradient value (alternatively, the peak gradient value) of the sequence of gradients as, for example, predetermined portion of the interpolated peak gradient value.
• the shaded region has four corners of interpolated gradients.
• the area divided by the interpolated peak gradient value is a good edge-sharpness measure, as it has a unit of length. It is noted that alternative definitions of the region are possible. For example, the region may be bounded from above not by the gradient level Li but by the sequence of gradients.
• FIG. 28 illustrates a sequence of gradients of samples of the image data across an edge plotted against distance in multiples of a spacing between successive gradients, a center of gravity 3401 (i.e. center of moment), and distances u ⁇ , U 3 , U 4 , U 5 and ue of the gradients (having gradient values g ⁇ , gs, g 4 , g 5 and ge) from the center of gravity.
• a good edge- sharpness measure is a if-th central moment of the gradients about the center of gravity, namely a weighted average of the distances of the gradients from the center of gravity with the weights being magnitudes of the respective gradients, k being an even integer.
• k can be 2, which makes the edge-sharpness measure a variance as if the sequence of gradients were a probability distribution.
• the edge-sharpness measure has a unit of a square of a unit of length. More generally, the edge-sharpness measure may be a function of distances of a plurality of gradients of a
• the predefined position may be an interpolated peak position for the sequence of gradients.
• a proper subset of the gradients of edge may be chosen according to a predefined criterion to participate in this calculation.
• the gradients may be required to have gradient values at least a predetermined fraction of the peak gradient or gradient value of an interpolated peak of the sequence of gradients.
• FIG. 25 illustrates a sequence of second derivatives of a sequence of samples of image data across an edge plotted against distance in multiples of a spacing between successive second derivatives, showing (a) a width W s between a pair of positive and negative peaks, (b) a width Wi between a pair of outermost interpolated second derivatives that have a given magnitude h lr (c) a width i3 ⁇ 4> between an inner pair of
• interpolated second derivatives that have the given magnitude h lr and (d) a distance Di from a zero-crossing (between the pair of positive and negative peaks) to an outermost
• any one of the three widths W s , Wi and i3 ⁇ 4> may used as the edge- sharpness measure.
• the edge- sharpness measure may be a weighted sum of distances from the zero-crossing (between the pair of positive and negative peaks, and may be interpolated) of the second derivatives with the weights being magnitudes of the respective second
• the edge-sharpness measure may be a function of distances of a plurality of second derivatives across the edge from a predefined position relative to the plurality of second derivatives.
• a center of gravity is a good candidate for the predefined position, with the weights being magnitudes of the second derivatives.
• Yet another good candidate for the edge-sharpness measure may be a function of distances of a plurality of second derivatives across the edge from a predefined position relative to the plurality of second derivatives.
• a center of gravity is a good candidate for the predefined position, with the weights being magnitudes of the second derivatives.
• Yet another good candidate for the edge-sharpness measure may be a function of distances of a plurality of second derivatives across the edge from a predefined position relative to the plurality of second derivatives.
• a center of gravity is a good candidate for the predefined position, with the weights being magnitudes of the second derivatives.
• Yet another good candidate for the edge-sharpness measure
• predefined position may be the midway point between the pair of positive and negative gradients.
• FIG. 26 illustrates a sequence of samples of image data from pixels of an edge plotted against distance in multiples of a spacing between contiguous pixels, showing (a) a width W edge and a contrast C edge between two samples at two ends of the edge, (b) a peak gradient value g peak (generated by the Roberts operator) between a pair of samples that has a steepest change of sample value, (c) a narrowest undivided portion of the edge that has contrast Ci and width W partl , and (d) a narrowest undivided portion of the edge that has contrast C ⁇ and width W part 2-
• the peak gradient value g pea k divided by the contrast C edge is a good edge-sharpness measure.
• the width W edge is another good edge-sharpness measure.
• the widths W part i and W part 2 are also good alternatives.
• contrasts Ci and/or C ⁇ may be defined to be a predetermine portion of the edge contrast C eC fge- I Alternatively, any one of them may be defined to be a predetermined multiple of a peak gradient of the edge, such as the peak gradient g pea k- It is also noted here that the "narrowest undivided portion" may be delimited by interpolated samples of image data, such as shown in squares in Figure 26, or by rounding down or up to a nearest pixel count.
• Figure 23A and Figure 23B illustrate a method where the focus signal generator detects a lack of symmetry about a peak in a gradient signal (also referred to below as gradient profile) to de-emphasize or eliminate
• a gradient signal also referred to below as gradient profile
• the peak may be a peak gradient among a series of consecutive gradients.
• the peak may be an interpolated peak gradient that is interpolated from two or more gradients among the series of
• Figure 23A illustrates a gradient profile of an image signal across two adjacent edges of opposite polarities (i.e. signs) where the edges are apart and do not interact.
• Figure 23B illustrates a gradient profile of an image signal across two adjacent edges of opposite polarities where the edges are close enough to mutually interact. It is clear from comparing Figures 23A and 23B that where adjacent edges of opposite signs (i.e. one of the edges has positive gradients, while the other one has negative gradients) become close, their respective gradient profile loses symmetry .
• the gradient profile Adjacent to each peak gradient 3210, 3260, respectively, the gradient profile has a left-to-right symmetry about the peak.
• the positive gradient profile 3211 and the negative gradient profile 3261 each
• a positive peak gradient 3212 on the left at position 6 and a negative peak 3262 on the right at position 9 are closer together than in Figure 23A.
• the gradient values are normalized to give peak gradient magnitudes of 1.0.
• the edges that correspond to the positive gradient profile 3213 and negative gradient profile 3263, respectively, in Figure 23B apparently interact to partially cancel each other, causing a reduction in magnitudes of gradients that lie between the closely adjacent positive 3212 and negative 3262 peaks at positions 6 and 9, respectively.
• the gradient profiles 3213, 3263 lack left-right reflection symmetry over the respective edges.
• the lack of symmetry is particularly salient in the interpolated gradient profile (shown in solid curve) in the figures.
• the lack of symmetry can be found within a certain distance from the peak gradient (in particular, the distance between 0.2 to 0.7 times the edge width of the edge) or a certain range of gradient levels between its peak gradient level and a non-zero fraction thereof (in particular, within 10% to 90% of the peak gradient level; more particularly, within 20% to 80%) .
• asymmetry may be found by comparing an the left side and the right side of the interpolated gradient profile within a distance of half the edge width, or alternatively within a range of gradient levels between 20% and 80% of its peak gradient level.
• edge width around each peak 3212, 3262 is reduced to 4, measured using the same threshold of 0.3 times the peak gradient magnitude.
• the edge widths of both edges thus measured no longer
• generator may detect the asymmetry and either cause a de- emphasis of a contribution of an associated edge width towards a focus signal and/or an edge count or an
• sharpness of edges in the images may de-emphasize or eliminate altogether an influence of an edge across which the gradient profile lacks symmetry.
• Isolated edges in an image that arise from sharp boundaries in the scene have gradient profiles that each reveals a left-right reflection symmetry across the respective isolated edge.
• FIG 23A there is a left- right reflection symmetry along the vertical symmetry axis (vertical dashed line) , which happens to be under the peak gradient 3210 at position 6 such that under a reflection along the vertical axis of symmetry the gradient at position 4 is mapped to the gradient at position 7 and vice versa, the gradient at position 3 is mapped to the gradient at position 8 and vice versa, and so on.
• Figure 24J illustrates another typical gradient profile of an isolated edge.
• An interpolated peak 3270' is shown at position 5.85 approximately, and a vertical axis of symmetry 3271' is shown in dash-dot line under the interpolated peak 3270'.
• each gradient in the gradient profile does not map to another gradient in the same gradient profile under a reflection along the axis of symmetry 3271', it does map to an interpolated gradient.
• the gradient at position 3 maps to an interpolated gradient marked with "X" (at
• interpolated gradient maps to another interpolated gradient and vice versa such as the pair of interpolated gradients marked with a triangle and an inverted
• Such geometries include point (i.e. gradient or interpolated gradient) , line, curve, bounded region, corner, etc. Corresponding to such geometries are
• -A ⁇ X-Y ⁇ B where A and B are positive numbers, such that values of X-Y greater than -A and less than B do not result in a determination of asymmetry, whereas values of X-Y either more positive than B or more negative than -A will result in determination of excessive lack of symmetry.
• the range of values of X-Y less negative than -A and less positive than B is referred to hereinafter as tolerance region, and the limits of the tolerance region are the asymmetry thresholds.
• -A and B are both asymmetry thresholds that delimit the tolerance region for the asymmetry that X and Y measures.
• exceeding the (relevant) asymmetry threshold For example, if (X-Y) is more
• This aspect of the invention i.e. eliminating or attenuation a contribution of an edge towards a focus signal or towards a focus control, is not limited to the specific methods discussed below for detecting lack of reflection symmetry of the gradient profile across the edge but include their equivalents, approximations, obvious or known variations, as well as include any computational method that makes use of one or more of the properties of reflection symmetry discussed above.
• One method to detect the lack of symmetry is to find a difference between a count of pixels on a side of the peak gradient and that on the other side, gradient magnitudes associated with the pixels being above a certain fraction of a peak gradient magnitude.
• a count asymmetry threshold may be set at, for example 0.5, so that when any one side counts more pixels than the other side in excess of the count asymmetry threshold, lack of symmetry is detected. This is illustrated by way of an example using Figure 23A and Figure 23B and a fraction of 0.3 and a count asymmetry threshold of 0.5.
• Figure 23A and Figure 23B and a fraction of 0.3 and a count asymmetry threshold of 0.5.
• a modification is to interpolate from the gradients to find a fractional pixel position 3272 ("interpolated peak position") where an interpolated gradient profile attains a maximal magnitude ("interpolated peak position")
• This interpolated peak position may be used to calculate the distances to the left and to the right as described below.
• the interpolated peak gradient may also be used to calculate the gradient level at which those distances are measured or above/below which pixels are counted. For example, in Figure 24A, a vertical dash- dot line is drawn under an interpolated peak 3270, a horizontal dotted line 3275 is drawn across the
• a modification to the above method is to determine the distances to the left and the right, respectively, from the peak gradient to where the gradient profile is interpolated to cross a certain gradient level that is a fraction (preferably between 10% and 90%, more preferably between 20% and 80%) (the "crossings") of the gradient value of the peak gradient 3212 (alternatively, the interpolated peak 3270), and find a lack of symmetry if the larger distance exceeds the smaller distance by a certain width asymmetry threshold or more. In other words, one distance subtracts the other distance being more negative than - (width asymmetry threshold) or more positive than the width asymmetry threshold will cause a determination of lack of symmetry.
• the tolerance region thus occupies an interval of number symmetrical about zero.
• the width asymmetry threshold may be determined in one of several ways. It may be given as a fixed number for an image, or a number that depends on the edge width of the edge associated with the peak, such as 10% of the edge width if the edge width is 3 or less, and 7% of the edge width if the edge width is wider than 3 but less than 5. Other reasonable dependencies based on how the image signal (from which the gradients in the gradient profile are generated) and/or how the gradients in the gradient profile are generated are acceptable for
• Figure 24A also illustrates this asymmetry detection method.
• the distances may be measured from the peak gradient 3212 (at position 6) , or alternatively from the interpolated peak 3270 (at approximately position 5.8).
• the distances W L and W R are measured from the interpolated peak 3270, giving approximately 2.5 and 1.3,
• the edge width is measured at normalized gradient level of +0.3, giving approximately 3.7.
• the width asymmetry threshold may be given as a fraction of the edge width, for
• An alternative method is to evaluate two areas, one to the left and the other to the right of the peak gradient 3212 (alternatively the interpolated peak 3270), and compare them according to a prescribed criterion against an area asymmetry threshold.
• Each of the two areas may be bounded on one side by a vertical line below the peak gradient (or the interpolated peak) , on the other side by the interpolated gradient (in solid curve) (or, alternatively, straight lines connecting consecutive gradients) , and from the top and bottom by an upper gradient level and a lower gradient level each at a different predetermined fraction of the peak gradient level (or, alternatively, interpolated peak gradient level, i.e. gradient level of the interpolated peak) (alternatively, no upper gradient level limits the area but just the gradients or interpolated gradient profile) .
• an upper gradient level 3276 is drawn at 0.75 and a lower gradient level 3274 at 0.2.
• a region 3277 (with area A L ) (left of the positive interpolated peak 3270) is bounded from above by the upper gradient level 3276, from below by the lower gradient level 3274, from the right by the vertical dash-dot line under the interpolated peak, and from the left by the interpolated gradient profile (solid curve) .
• a region 3278 (having area A R ) (right of the same peak 3270) is similarly bounded from above and below, and is bounded from the right by the interpolated gradient profile and from the left by the vertical dash-dot line. A lack of symmetry is detected when the areas A L and A R differ beyond a
• the asymmetry may be detected when the larger area exceeds the smaller area by an area asymmetry threshold or more.
• the area asymmetry threshold may be expressed in one of various different ways. It may be expressed in terms of a percentage (of the lesser area) , which may be a fixed number for the image or, alternatively, a function of the edge width of the associated edge. Alternatively, it may be expressed in terms of an area difference for the normalized gradient profile. Other reasonable dependencies based on how the image signal (from which the gradients in the gradient profile are generated) and/or how the gradients in the gradient profile are generated are acceptable for
• a common distance W 0 is measured from the interpolated peak 3270 (or, alternatively, peak gradient 3212) to the left and right sides of the gradient profile.
• interpolated gradients are calculated (or gradient is found) such that their distances from the vertical dash- dot line under the interpolated peak 3270 (or peak gradient 3212) are both W 0 .
• both interpolated gradients would be at a common gradient level.
• the interpolated gradients lie on different gradient levels G 2 3252, G h 3253.
• a lack of symmetry is detected when the gradient levels G 2 and G h differ beyond a predetermined limit according to a prescribed criterion.
• the asymmetry may be detected when the excess G hl of the larger gradient level G h 3253 over the smaller gradient level G 2 3252 exceeds the smaller gradient level G 2 by an gradient asymmetry threshold or more.
• the gradient asymmetry threshold may be expressed in one of various different ways. It may be expressed in terms of a percentage (e.g. of the lesser gradient G 2 ) , which may be a fixed number for the image or, alternatively, a function of the edge width of the associated edge. Alternatively, it may be expressed in terms of a gradient level difference for the normalized gradient profile.
• the common W 0 may be selected to be a predetermine fraction of the edge width, such as a fraction between 0.1 and 0.5, preferably between 0.2 and 0.4.
• W 0 may be selected as the lesser of the two distances from the interpolated peak 3270 (or, alternatively, the peak gradient 3212) to a pair of interpolated gradients or gradients at a given gradient level that is a
• G h alone can be the parameter to indicate a degree of asymmetry.
• a gradient asymmetry threshold then may be set such that when G h exceeds the threshold the lack of asymmetry is detected.
• a modification of the immediate above method is to compare between first or second derivatives at those two interpolated gradients at gradient levels W 1 and W h , respectively.
• both interpolated gradients would be have first derivatives that are opposite in their signs but
• the interpolated gradients usually differ in first and second derivatives.
• a lack of symmetry is detected when the magnitude of the first derivative differs between the two interpolated gradients (or possibly gradients) beyond a predetermined limit according to a prescribed criterion.
• the asymmetry may be detected when the larger first
• the asymmetry threshold may be expressed in one of various different ways. It may be expressed in terms of a
• a function of the edge width of the associated edge is acceptable for determining the gradient asymmetry threshold.
• a vertical line can be drawn from a midpoint between the pair of intersections between the upper gradient level
• the interpolated gradient curve has a segment on the left (having a length L L ) between normalized gradient levels of 0.25 and 0.75, longer than a segment on the right, whose length L R is clearly shorter, indicating a lack of symmetry.
• a lack of symmetry is detected when the lengths L L and L R differ beyond a predetermined limit according to a prescribed criterion.
• the asymmetry may be detected when the longer length exceeds the shorter length by a length asymmetry threshold or more.
• the length asymmetry threshold may be expressed in one of various different ways. It may be expressed in terms of a
• percentage (such as of the lesser length), preferably from 10% to 30%, which may be a fixed number for the image or, alternatively, a function of the edge width of the associated edge. Alternatively, it may be expressed in terms of a length difference for the normalized gradient profile. Other reasonable dependencies based on how the image signal (from which the gradients are generated) and/or how the gradients are generated are acceptable for determining the length asymmetry
• Figure 24C may be modified. Instead of comparing lengths L L and L R , the areas A' L and A' R of the shaded regions on the left and right sides, respectively, can be compared in a similar manner.
• An area asymmetry threshold may be defined similarly and used to compare with a magnitude of difference between A r L and A r R .
• the length method described immediately above and illustrated using Figure 24C may be modified in yet another way.
• the distance on the left (W BL ) between where the interpolated gradient curve intersects the upper and lower gradients, respectively, with that on the right (W BR ) is compared.
• the a lack of symmetry is found if W BL and W BR differ too much according to a prescribed
• a lack of symmetry is found when the larger of W BL and W BR exceeds the smaller one by more than a width asymmetry threshold.
• the width asymmetry threshold may be prescribed in a manner like any one of the various asymmetry thresholds above.
• Figure 24H illustrates an alternative method that is equivalent to the width method described immediately above. This method calculates a first midpoint 3281
• inter- midpoint distance a distance between the first 3281 and second 3280 midpoints
• a lack of symmetry is detected when the inter-midpoint distance exceeds a certain inter-midpoint- distance asymmetry threshold.
• the inter-midpoint distance is twice of ⁇ W BL - W BR ⁇ . In a variation on this method, only one gradient level 3274 is used and only the
• a distance X bPk is measured from the peak gradient 3212 (or,
• Yet another method is to find first derivatives of slopes of the gradient profile on two sides of the peak gradient 3212 (alternatively, the interpolated peak 3270) and compare the first derivatives under a prescribed criterion to determine whether there is a lack of
• first derivatives on both sides that only differ in sign but are identical in magnitude.
• the first derivatives may be calculated approximately by an
• the first derivatives may be calculated approximately at a gradient level that is a certain fraction (preferably between 10% and 90%, more preferably between 20% and 80%) of the peak gradient value, for example 0.5.
• a gradient level that is a certain fraction (preferably between 10% and 90%, more preferably between 20% and 80%) of the peak gradient value, for example 0.5.
• the gradient level is at 0.5 times the gradient level of the peak gradient 3212
• the right-side slope is conspicuously steeper than the left-side slope.
• Figure 24D shows how the first derivatives are evaluated on the normalized gradient profile at a gradient level of 0.25 and approximated with hypotenuses of right-angled triangles (shaded) that have base width of 1 touching the left and right sides of the interpolated gradient profile, respectively.
• the first derivatives are approximated by the heights of triangle S and S R , respectively.
• Figure 24E illustrates another way to approximate the first derivatives using two triangles that have identical base width (1 in this illustration) .
• the base of each triangle is centered at a position where the gradient profile is interpolated to be at the gradient level of 0.25.
• Each end of the corresponding hypotenuse is a half pixel away and take as gradient value a gradient value interpolated from the gradient profile.
• the first derivatives are approximated as the heights S' L and S' R divided by the base width, which is 1 in this illustration.
• there are various methods to approximate a first derivative from a sequence of data points and therefore this aspect of the invention is not limited to the particular examples given above but includes all equivalent methods and all
• Still another method is to find second derivatives of the gradient profile on two sides of the peak gradient 3212 (alternatively, the interpolated peak 3270) and compare the second derivatives under a prescribed
• the second derivatives may be calculated at a gradient level that is a certain fraction
• Figure 24F illustrates a method of how the second derivate may be approximated.
• Figure 24G illustrates another approximation for the second derivative, and at normalized gradient level of 0.18. On each side, one triangle is fitted to the
• each triangle has its hypotenuse inclined at an inclination to fit the hypotenuse to the interpolated gradient profile.
• the heights S UL and S LL of the triangles on the left (S m and S LR of the triangles on the right) are subtracted to find second derivative D L (D R ) .
• the opposite signs of the second derivates are indicated with the arrows: pointing up for D' L , pointing down for D' R .
• This gradient profile clearly has significant mismatch of second derivatives on two sides, thus is asymmetrical.
• asymmetry threshold or more generally a predetermined tolerance region such that when a value of the comparison parameter is outside the tolerance region the lack of reflection symmetry is detected.
• the comparison parameter is outside the tolerance region the lack of reflection symmetry is detected.
• distances between midpoints X ab ( Figure 24H) and the difference between first derivatives S R - S L ( Figure 24D) may be combined in a weighted average Z and then compared with an asymmetry threshold a that defines a tolerance region as an interval - a ⁇ Z ⁇ a.
• Detecting that the edge has asymmetry in a sequence of gradients across itself need not necessarily involve computation directly on the sequence of gradients.
• the asymmetry can be detected in the image sample values across the edge. Referring to Figure 26, for the two narrowest undivided portions of the edge that have contrasts CI and C2 respectivey, their centers will match if there is perfect symmetry in the gradients.
• misalignment between the centers indicates asymmetry.
• the misalignment can be measured and divided by a width of the edge to provide a quantity that indicates asymmetry in the gradient profile across the edge.
• a distance from the center of one of the undivided portions corresponding to contrast CI to where the sequence of gradient has the steepest rise or fall with peak gradient g peak (according to Roberts detector) and divided by a width of the edge to indicate a degree of asymmetry.
• asymmetry can be measured from second derivatives of the image samples. Referring to Figure 25, the differences in the distances of the
• a quantity from an edge is said to be
• peak gradient 3212 has a
• length filter 212 creates a preference for edges that each connects to one or more edges of a similar orientation. A group of edges that are similarly oriented and mutually connected within the group
• concatenated edge is less likely to be due to noise, compared with an isolated edge that does not touch any other edge of similar orientation.
• the probability of the group being due to noise falls off exponentially as the number of edges within the group increases, and far faster than linearly.
• This property can be harnessed to reject noise, especially under dim- lit or short-exposure situations where the signal-to- noise ratio is weak, e.g. less than 10, within the image or within the region of interest.
• the preference may be implemented in any reasonable method to express such preference. The several ways described below are merely examples .
• a first method is to eliminate edges that belong to vertical/horizontal concatenated edges having lengths lesser than a concatenated length threshold.
• the concatenated length threshold may be larger when the region of interest is dimmer. For example, the
• concatenated length threshold may start as small as 2, but increases to 8 as a signal-to-noise ratio within the region of interest drops to 5.
• the concatenated length threshold may be provided by the processor 112, 112', 112", for example through a ⁇ length command' signal, shown in Figure 3, as part of signals 132. Alternately, the threshold may be calculated according to a formula on the focus signal generator.
• a second method is to provide a length-weight in the length filter 212 for each edge and apply the length- weight to a calculation of focus signal in the focus signal calculator 210. An edge that is part of a longer concatenated edge receives a larger weight than one that is part of a shorter concatenated edge.
• the length-weight may be a square of the length of the concatenated edge.
• a contribution of each edge towards the focus signal may be multiplied by a factor A/B before summing all contributions to form the focus signal, where B is a sum of the length-weights of all edges that enter the focus signal calculation, and A is a length-weight of the edge.
• the edge-width histogram which may be output as part of signals 134, may have edges that are members of longer concatenated edges contribute more to the bins corresponding to their respective edge width, thus preferred, instead of all edges contribute the same amount, e.g. +1.
• each edge may contribute A/C, where C is an average value of A across the edges.
• the narrow-edge count may have edges that are members to longer concatenated edges contribute more.
• the contribution from each edge may be
• A/D multiplied by A/D, where D is an average of A among edges that are counted in the narrow-edge count.
• D is an average of A among edges that are counted in the narrow-edge count.
• a group of N vertical (horizontal) edges where, with the exception of the top (leftmost) and the bottom
• Figure 8 illustrates a vertical concatenated edge and its length.
• cells R2C3 and R2C4 form a first vertical edge
• cells R3C3, R3C4, and R3C5 together form a second vertical edge
• cells R4C4 and R4C5 together form a third vertical edge.
• the first and the third vertical edges each touches only one other vertical edge
• the second vertical edge touches two other vertical edges.
• the first, second and third vertical edges together form a vertical concatenated edge having a length of 3.
• (horizontal) concatenated edge has two or more branches, i.e. having two edges in a row (column), the length may be defined as the total number of edges within the
• the length may be
• a definition of a length for a concatenated edge shall have a property that the length is proportional to the number of member edges within the concatenated edge at least up to three. This is to be consistent with the previously stated reasoning that more edges being
• the length filter 212 may de-emphasize or eliminate and thus, broadly speaking, discriminate against an edge having a
• the length filter 212 may discriminate against an edge having a concatenated length of two.
• the length filter 212 may discriminate against an edge having a concatenated length of three, to further reduce an influence of noise.
• the length filter 212 may do any one of these actions under a command from the processor.
• Filter 212 may be inserted before the focus signal calculator 210, wherein the edges processed by the Length Filter 212 are those that pass through the width filter 209 depending on the ⁇ fine' signal.
• the fine switch 220 may be removed so that focus signal calculation unit 210 receives a first set data not filtered by the width filter 209 and a second set filtered, and for each calculates a different focus signal, gross focus signal for the former, fine focus signal for the latter, and outputs both to the processor 112, 112' .
• FIG. 11 plots a histogram of edge widths, i.e. a graph of edge counts against edge widths.
• edge width of 2 i.e. the aforementioned sharp_edge_width
• there is a peak indicating a presence of sharp edges in the image.
• edge widths of 4 and 5 there are peaks, indicating edges that are blurred, possibly due to the corresponding imaged objects being out of focus, being at a different distance away from the focus lens than those objects that give rise to the sharp edges.
• edges whose widths lie outside a predetermined range (“narrow- edge range”) may be de-emphasized using the Width Filter 209.
• the Width Filter 209 may create a lesser weight for edge widths outside the narrow-edge range for use in the focus signal calculation.
• edge widths may be assigned weight of 1.0, whereas edges widths more than +1 to the right of the upper limit 840 assigned a weight of 0, and edge widths in between assigned weights between 0 and 1.0, falling monotonically with edge width.
• the Width Filter 209 may prevent such edges from entering the focus signal calculation altogether.
• Appropriate upper and lower limits 830, 840 depend on several factors, including crosstalk in the pixel array 108, the interpolation method used to generate missing colors for the image received by the focus signal
• upper and lower limits 830, 840 and the parameter sharp_edge_width may be determined for the image pickup apparatus 102, 102' by capturing images of various degrees of sharpness and inspecting the edge width histograms. For example, if a sharp image has a peak at edge width of 2, an appropriate lower and upper limit may be 1.5 and 3, respectively, and the sharp_edge_width may be set to 2.0.
• the lower and upper limits and sharp_edge_width may be determined as above and provided to the focus signal generator 120, 120', 120" by the processor 112, 112". When ⁇ fine command' is ON, the fine focus signal thus calculated de- emphasizes edge widths outside the narrow-edge range.
• the Width Filter 209 may calculate a total count of the edges whose edge widths fall within the narrow-edge range and output as part of output signals 134. Narrow-Edge Count may be input to and used by the focus system controller (processor 112) to detect a presence of sharp image and/or for initiating tracking.
• Focus Signal Referring next to the focus signal calculator 210 of
• the focus signal calculator 210 receives edge widths and outputs a focus signal.
• the weight at each edge width may be the edge count for the edge width multiplied by the edge width itself, i.e.
• Wi Ciei.
• the focus signal would be a value very close to the sharp edge width of 2.0 for the example shown in Figure 11, indicating that among object details within the focus distance range that would produce edge widths between 2.0 and 3.0, most are
• control signal ⁇ fine' is OFF and ⁇ exclude' is OFF
• the focus signal may be a value close to 5.0, indicating that there are substantial details of the image that are out of focus. Turning ON the fine switch 220 allows the focus signal to respond more to objects slightly blurred while less to those that are completely blurred.
• the fine switch 220 is ON, we shall refer to the focus signal as a fine focus signal, whereas when the fine switch 220 is OFF, a gross focus signal.
• the emphasis expressed by the Length Filter 212 may be incorporated into the focus signal in one of several ways, such as eliminating an edge that is de-emphasized from entering the focus signal calculation, or reducing a weight of the edge's contribution towards a count ei of a corresponding edge width bin.
• Figure 15 sketches a response of the fine focus signal to an adjustment of the focus position in the vicinity of where an object is in sharp focus.
• the fine focus signal reaches a minimum value, approximately at sharp_edge_width, where the focus position brings an image into sharp focus, and increases if otherwise.
• the fine focus signal may be used for tracking objects already in-focus or very nearly so. For moving objects, the fine focus signal allows the focus control system to keep the objects in sharp focus even if the focus
• Fine focus signal may also be used to acquire a sharp focus ("acquisition") of an object that is not yet in sharp focus but close enough such that the object gives rise to edges whose widths fall within the narrow-edge range. Since the edge width histogram exhibits a peak at the edge width corresponding to the object away from the sharp_edge_width, resulting in the fine focus signal being larger than the
• the focus control system may respond by adjusting the focus position to bring the fine focus signal value towards the sharp_edge_width, thus centering the peak of edge width due to the object at the edge width value equal to sharp_edge_width .
• Basic Use Figures 12-16 illustrate how the narrow-edge count, gross focus signal, and fine focus signal may be used to perform focus control to achieve sharp images.
• Figure 12 illustrates an outdoor scene having 3 groups of objects at different focus distances: "person” in the foreground, “mountain, sun, and horizon” in the background, and “car” in the between.
• Figure 13 is an illustration of the narrow-edge count plotted against time when the focus position of the focus lens 104 sweeps from far to near for the scene
• the narrow-edge count peaks when the focus position brings an object into a sharp image on the pixel array 108.
• the narrow-edge count plot exhibits 3 peaks, one each for "mountain, sun, and horizon", “car”, and “person”, in this order, during the sweep .
• Figure 14 shows the gross focus signal plotted against time.
• the gross focus signal exhibits a minimum when the focus position is near each of the 3 focus positions where the narrow-edge count peaks. However, at each minimum, the gross focus signal is not at the sharp edge width level, which is 2.0 in this example, due to bigger edge widths contributed by the other objects that are out-of-focus .
• Figure 15 illustrates the fine focus signal plotted against the focus position in the vicinity of the sharp focus position for "car” in the scene of Figure 12.
• the fine focus signal achieves essentially the sharp edge width, which is 2 in this example, despite the presence of blurred objects ("person” and "mountains, sun, and horizon”) .
• a focus control system may use the gross focus signal to search for the nearest sharp focus position in a search mode. It can move the focus position away from the current focus position to determine whether the gross focus signal increases or decreases. For example, if the gross focus signal increases (decreases) when the focus position moves inwards (outwards) , there is a sharp focus position farther from the current focus position.
• the processor 112, 112', 112" can then provide a focus drive signal to move the focus lens 104 in the direction
• a focus control system may use the fine focus signal to track an object already in sharp focus to maintain the corresponding image sharp (thus a "tracking mode")
• any shift in the fine focus signal level immediately informs the processor 112, 112', 112" of a change in the focus distance of the object.
• the processor 112, 112', 112" can then determine a
• the image pickup apparatus 102, 103, 103', 103" is able to track a moving object.
• a focus control system may use narrow- edge count to trigger a change from a search mode to a tracking mode.
• the focus control system uses the fine focus signal to "lock" the object.
• the focus control system may use the gross focus signal to identify the direction to move and regulate the speed of movement of the lens.
• narrow-edge count peaks sharply.
• the processor 112, 112', 112" may switch into the tracking mode and use the fine focus signal for focus position control upon detection of a sharp rise in the narrow-edge count or a peaking or both.
• a threshold which may be different for each different sharp focus position, may be assigned to each group of objects found from an end-to-end focus position "scan", and subsequently when the narrow-edge count surpasses this threshold the corresponding group of objects is detected.
• an end-to-end focus position scan can return a list of maximum counts, one maximum count for each peaking of the narrow-edge count.
• a list of thresholds may be generated from the list of maximum counts, for example by taking 50% of the maximum counts.
• Figure 16 illustrates an image pickup apparatus 102 having a display 114, an input device 107 comprising buttons, and selection marker 1920 highlighted in the display 114.
• a user can create, shape and maneuver the selection marker 1920 using input device 107.
• input device 107 may comprise a touch-screen overlaying the display 114 to detect positions of touches or strokes on the display 114.
• Input device 107 and processor 112, 112', 112" or a separate dedicated controller (not shown) for the input device 107 may determine the selection region.
• the parameters for describing the selection region may be transmitted to the focus signal generator 120, 120', 120" over bus 132 (or internally within the processor 112 in the case where focus signal generator 120 is part of the processor 112) .
• the focus signal generator 120 may limit the focus signal calculation or the narrow- edge count or both to edges within the selection region described by said parameters or de-emphasize edges outside the selection region. Doing so can de-emphasize unintended objects from the focus signal and then even the gross focus signal will exhibit a single minimum and a minimum level within 1.0 or less of the sharp edge width .
• FIG 17 shows an alternate embodiment of a focus signal generator 120'.
• Focus signal generator 120' outputs statistics of edges and edge widths.
• edge-width statistics that controller 120' outputs may be one or more of the following: an edge-width histogram comprising edge counts at different edge widths; an edge width where edge width count reaches maximum; a set of coefficients representing a spline function that approximates edge counts at different edge widths; and any data that can represent a function of edge width.
• Census Unit 240 may receive data computed in one or more of the other units with the focus signal generator 120' to calculate statistics of edge widths.
• the focus signal generator 120' may output a signal that has an indication of a distribution of edge widths.
• the edge-width statistics thus provided in signals 134 to an alternative embodiment of processor 112' in an alternative auto-focus image pickup apparatus 102' may be used by the processor 112' to compute a gross and/or fine focus signal and a narrow- edge count in accordance with methods discussed above or equivalent thereof.
• any data computed in the focus signal generator 120' may be output to the processor 112' as part of the output signals 134.
• the processor 112' may internally generate a focus signal and/or a narrow-edge count in addition to the functions included in the processor 112 of Figure 1.
• interpolator 148, and generator 120' may reside within a package 142, together comprising an image sensor 150', separate from the processor 112'.
• Auxiliary Pixel Array Figure 19 shows an alternate embodiment of an auto- focus image pickup system 103.
• the system 103 may include a partial mirror 2850, a full mirror 2852, an optical lowpass filter 2840, a main pixel array 2808, and a main A/D Converter 2810.
• the partial mirror 2850 may split the incoming light beam into a first split beam and a second split beam, one transmitted, the other reflected.
• the first split beam may further pass through the optical lowpass filter 2840 before finally reaching the main pixel array 2808, which detects the first split beam and converts to analog signals.
• the second split beam may be reflected by the full mirror 2852 before finally reaching the auxiliary pixel array 108", which corresponds to the pixel array 108 in system 102 shown in Figure 1.
• the ratio of light intensity of the first beam to the second beam may be 1-to-l or greater than 1-to-l.
• the ratio may be 4-to-l.
• the main pixel array 2808 may be covered by a color filter array of a color mosaic pattern, e.g. the Bayer pattern.
• the optical lowpass filter 2808 prevents the smallest light spot focused on the pixel array 2808 from being too small as to cause aliasing.
• aliasing can give rise to color moire artifacts after a color interpolation, .
• the smallest diameter of a circle encircling 84% of the visible light power of a light spot on the main pixel array 2808 (“smallest main diameter") may be kept larger than one and a half pixel width but less than two pixel widths by use of the optical lowpass filter.
• the optical lowpass filter 2840 may be
• the auxiliary pixel array 108" may comprise one or more arrays of photodetectors . Each of the arrays may or may not be covered by a color filter array of a color mosaic pattern.
• the array (s) in auxiliary pixel array 108" outputs image (s) in analog signals that are
• A/D Converter 110 converts digital signals 130 by A/D Converter 110 to digital signals 130 by A/D Converter 110.
• the images are sent to the focus signal generator 120.
• a color interpolator 148 may generate the missing colors for images generated from pixels covered by color
• auxiliary pixel array 108 comprises
• each array may capture a sub-image that corresponds to a portion of the image captured by the main pixel array 2808.
• the multiple arrays may be physically apart by more than a hundred pixel widths, and may or may not share a semiconductor substrate. Where the pixel arrays within auxiliary pixel array 108" do not share a semiconductor substrate, they may be housed together in a package (not shown) .
• Main A/D Converter 2810 converts analog signals from the Main Pixel Array 2808 into digital main image data signal 2830, which is sent to the processor 112, where the image captured on the Main Pixel Array 2808 may receive image processing such as color interpolation, color correction, and image compression/decompression and finally be stored in memory card 116.
• An array of photodetectors in the auxiliary pixel array 108" may have a pixel width ("auxiliary pixel width") that is smaller than a pixel width of the main pixel array 2808 ("main pixel width”) .
• the auxiliary pixel width may be as small as half of the main pixel width. If an auxiliary pixel is covered by a color filter and the auxiliary pixel width is less than 1.3 times the smallest spot of visible light without optical lowpass filtering, a second optical lowpass filter may be inserted in front of the auxiliary array 108" to increase the smallest diameter on the auxiliary pixel array 108" ("smallest auxiliary diameter") to between 1.3 to 2 times as large but still smaller than the smallest main
• the slight moire in the auxiliary image is not an issue as the auxiliary image is not presented to the user as the final captured image.
• Figure 22 illustrates how edge widths may vary about a sharp focus position for main images from the main pixel array 2808 (solid curve) and auxiliary images from the auxiliary pixel array 108" (dashed curve) .
• the auxiliary images give sharper slopes even as the main images reach the targeted sharp edge width of 2.
• the auxiliary image is permitted to reach below the targeted sharp edge width, since moire due to aliasing is not as critical in the auxiliary image, as it is not presented to the user as a final image. This helps to sharpen the slope below and above the sharp edge width.
• the sharper slope is also helped by the auxiliary pixel width being smaller than the main pixel width.
• the shaded region in Figure 22 indicates a good region within which to control the focus position to keep the main image in sharp focus. A change in focus
• a linear feedback control system may be employed to target the middle auxiliary edge width value within the shade region and to use as feedback signal the edge widths generated from the auxiliary images.
• the auxiliary pixel array 108", A/D Converter 110, focus signal generator 120 together may be housed in a package 142 and constitute an auxiliary sensor 150.
• the auxiliary sensor 150 may further comprise a color
• Figure 20 shows an alternative embodiment of auto- focus image pickup apparatus 103' similar to apparatus 103 except focus signal generator 120' replaces focus signal generator 120.
• the auxiliary pixel array 108", A/D Converter 110, focus signal generator 120' together may be housed in a package 142 and constitute an
• the auxiliary sensor 150 may further comprise a color interpolator 148.
• Figure 21 shows an alternate embodiment of auto-focus image pickup apparatus 103".
• the focus signal generator 120 and the processor 112" may be housed in a package 144 as a camera controller, separate from the auxiliary pixel array 108".
• the processor 112" is similar to processor 112 except that processor 112" receives images from the main pixel array 2808 as well as the auxiliary pixel array 108".
• the processor 112" may perform a color interpolation, a color correction, a
• the processor 112" may perform color interpolation on images received on signal 130 for pixels that are covered by color filters in the auxiliary pixel array 108" and send the color interpolated images to the focus signal generator 120 on signal 146.
• the auto-focus image pickup system 102, 102', 103, 103' , 103" may include a computer program storage medium (not shown) that comprises instructions that causes the processor 112, 112', 112" respectively, and/or the focus signal generator 120, 120' to perform one or more of the functions described herein.
• the instructions may cause the processor 112 or the generator 120' to perform a slant correction for an edge width in accordance with the flowchart of Figure 7.
• the instructions may cause the processor 112' or the generator 120 to perform an edge width filtering in accordance with the above description for Width Filter 209. Alternately, the processor 112, 112' or the
• generator 120, 120' may be configured to have a
• FIG. 30 shows yet another embodiment of focus signal generator 120'. This embodiment may be employed in any of the above image capture systems.
• any nonvolatile storage medium may be used instead, e.g. hard disk drive, wherein images stored therein are accessible by a user and may be copied to a different location outside and away from the system 102.
• One or more parameters for use in the system may be stored in a non ⁇ volatile memory in a device within the system.
• the device may be a flash memory device, the processor, or the image sensor, or the focus signal generator as a separate device from those.
• One or more formulae for use in the system for example for calculating the sharp_edge_width
• concatenated length threshold, or for calculating beta may likewise be stored as parameters or as computer- executable instructions in a non-volatile memory in one or more of those devices. While certain exemplary embodiments have been

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• General Physics & Mathematics (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Optics & Photonics (AREA)
• Studio Devices (AREA)
• Automatic Focus Adjustment (AREA)
• Focusing (AREA)
• Image Analysis (AREA)
• Image Processing (AREA)

Priority Applications (13)

Applications Claiming Priority (2)

Related Parent Applications (1)

Related Child Applications (1)

Publications (1)

Family

ID=44511103

Family Applications (1)

Country Status (8)

Cited By (2)

Citations (3)

Family Cites Families (4)

• 2011
  • 2011-06-09 MX MX2013006516A patent/MX2013006516A/es active IP Right Grant
  • 2011-06-09 AU AU2011340208A patent/AU2011340208A1/en not_active Abandoned
  • 2011-06-09 EP EP11748716.5A patent/EP2649788A1/en not_active Ceased
  • 2011-06-09 JP JP2013542633A patent/JP6057086B2/ja not_active Expired - Fee Related
  • 2011-06-09 DE DE112011104233T patent/DE112011104233T5/de not_active Withdrawn
  • 2011-06-09 WO PCT/IB2011/052529 patent/WO2012076993A1/en active Application Filing
  • 2011-06-09 CA CA2820847A patent/CA2820847A1/en not_active Abandoned
  • 2011-06-09 SG SG2013044219A patent/SG190755A1/en unknown

Patent Citations (3)

Non-Patent Citations (2)

Also Published As

Similar Documents

Legal Events