GB2589362A - Method and apparatus for processing image sensor output to improve image quality - Google Patents
Method and apparatus for processing image sensor output to improve image quality Download PDFInfo
- Publication number
- GB2589362A GB2589362A GB1917353.3A GB201917353A GB2589362A GB 2589362 A GB2589362 A GB 2589362A GB 201917353 A GB201917353 A GB 201917353A GB 2589362 A GB2589362 A GB 2589362A
- Authority
- GB
- United Kingdom
- Prior art keywords
- reference value
- baseline
- value
- pixel
- pixel values
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 16
- 230000001419 dependent effect Effects 0.000 claims 3
- GJJFMKBJSRMPLA-HIFRSBDPSA-N (1R,2S)-2-(aminomethyl)-N,N-diethyl-1-phenyl-1-cyclopropanecarboxamide Chemical compound C=1C=CC=CC=1[C@@]1(C(=O)N(CC)CC)C[C@@H]1CN GJJFMKBJSRMPLA-HIFRSBDPSA-N 0.000 abstract 1
- 230000008447 perception Effects 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000010348 incorporation Methods 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 229910044991 metal oxide Inorganic materials 0.000 description 1
- 150000004706 metal oxides Chemical class 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/60—Noise processing, e.g. detecting, correcting, reducing or removing noise
- H04N25/63—Noise processing, e.g. detecting, correcting, reducing or removing noise applied to dark current
- H04N25/633—Noise processing, e.g. detecting, correcting, reducing or removing noise applied to dark current by using optical black pixels
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
Abstract
A method of pixel processing comprises: subtracting a baseline value from a ixel value to produce a calibrated pixel value; selectively adjusting a baseline reference value to produce the baseline value; and, in respect of a plurality of pixel values, adjusting the baseline reference value for some of the pixel values and not adjusting the baseline reference value for the rest of the pixel values. A random or pseudo-random determination may be made regarding for which of the pixel values the baseline reference value is adjusted, e.g. using comparison of a random number with a comparator reference value, which may have shuffled bits. Adjustment of the baseline reference value may comprise adding one to its value, and the baseline reference value may comprise an integer or fractional part of a dark reference value. The plural pixel values may correspond to a row; baseline-adjusted pixels may be interspersed in the row that are not adjusted. Dark reference values may be fixed, user-adjustable, user-selected, or calculated from one or more dark reference pixel values.
Description
Method and Apparatus for Processing Image Sensor Output to Improve Image Qualify
Field of the Invention
The invention relates to image sensors. The invention relates particularly to processing image sensor output to improve image sensor quality, especially to compensate for variations caused by image sensor hardware and operating conditions.
Background to the Invention
Image sensors are well known and are typically implemented using charge-coupled device (CCD), complementary metal-oxide semiconductor (CMOS) or more recently scientific CMOS (sCMOS) technology. An image sensor comprises an array of pixels and associated hardware in the form of readout circuitry for outputting a pixel value for each pixel of the array. In order to calibrate the pixel values used to create an image the array may include dark reference pixels, the dark reference pixel values being used to calibrate the image pixel values. However, the readout circuitry components are not the same for each pixel and variations in the circuitry components can result in undesirable variations in the output pixel values.
For example, in an sCMOS image sensor each row of pixels is associated with its own readout circuit comprising an amplifier. Typically there are small variations in the operating characteristics of the respective amplifier for each row which can lead to variations in pixel values from row to row from a uniform light source (including darkness). In order to calibrate the image pixel values with respect to a baseline value (which is intended to represent true darkness as a zero pixel value), an average of the dark reference pixel values of the respective row may be subtracted from each image pixel value. Since the dark reference pixels of each row are processed by the same row amplifier as the image pixels of the respective row, the row-to-row amplifier related variation should be eliminated, giving a uniform row baseline across the entire image. This should allow the image pixel values to be scaled up without distortion caused by correspondingly exaggerating any row baseline differences.
However, it is found that row-to-row variations still exist and can give rise to row-based patterns in the resulting image that are discernible to the human eye. For example, such patterns can cause a perception that areas of darkness in the image are noisy and that the image quality is sub-optimum.
It would be desirable to mitigate the problems outlined above and in particular to compensate the 35 image sensor output for variations in the associated processing circuitry.
Summary of the Invention
A first aspect of the invention provides a pixel processing device comprising at least one instance of 40 a pixel processor, each instance of pixel processor comprising: a subtractor configured to subtract a baseline value from a pixel value to produce a calibrated pixel value; a baseline generator configured to provide said baseline value to said subtractor, said baseline generator comprising: an adjustor for selectively adjusting a baseline reference value to produce said baseline value; and means for determining whether or not to cause said adjustor to adjust said baseline reference value to produce said baseline value, wherein, in respect of a plurality of pixel values, said determining means is configured to cause said 10 adjustor to adjust said baseline reference value for some of said pixel values and not to adjust said baseline reference value for the rest of said pixel values.
Preferably, said determining means is configured to determine randomly, or pseudo-randomly, for which of said pixel values said baseline reference value is adjusted. In preferred embodiments, said determining means comprises a random number generator that is operable to output a random number for each of said pixel values, and a comparator for comparing each random number with a comparator reference value, and wherein for each of said pixels, said determining means is configured to cause said adjustor to adjust said baseline reference value or not depending on the comparison of the respective random number with the comparator reference value for the respective pixel value. Said random number generator may have a seed input from which random numbers are generated, and wherein said seed input is provided by said comparator reference value. Said comparator reference value typically comprises data bits, and said baseline generator may include a bit shuffling device for shuffling the bits of said comparator reference value.
In preferred embodiments, adjusting said baseline reference value comprises adding 1 to said baseline reference value.
Preferably, said baseline reference value comprises an integer part of a dark reference value. said comparator reference value preferably comprises a fractional part of a dark reference value.
In typical embodiments, said plurality of pixel values correspond to a row of pixels, and wherein said determining means is configured to cause the pixel values in respect of which the baseline reference value is adjusted to correspond with pixels in said row that are interspersed with other pixels in said row that correspond with pixel values in respect of which the baseline reference value is not adjusted. The determining means may be configured to cause the pixel values in respect of which the baseline reference value is adjusted to correspond with pixels in said row that are randomly or pseudo-randomly interspersed with other pixels in said row that correspond with pixel values in respect of which the baseline reference value is not adjusted.
From a second aspect the invention provides an image sensor system comprising an image sensor comprising an array of pixels, a pixel processing device of the first aspect of the invention, and readout circuitry for obtaining pixel values from said image sensor and providing said pixel values to said pixel processing device. Said plurality of pixel values may comprise image pixel values from a row of said array of pixels. In preferred embodiments, aid array comprises a plurality of dark reference pixels, and wherein said baseline reference value is calculated from, or otherwise derived from, the values of at least some of said dark reference pixels. Each row of said array may include a plurality of dark reference pixels, and wherein said baseline reference value comprises an integer part of a dark reference value calculated from, or otherwise derived from, the dark reference pixel values of the respective row. Said dark reference value may be an average of the dark reference pixel values of the respective row.
In preferred embodiments, the baseline reference value comprises a dark reference value, the dark reference value comprising either a fixed value, a user-adjustable value, a user-selected value or a value calculated from one or more dark reference pixel values.
A third aspect of the invention provides a digital camera comprising an image sensor system according to the second aspect of the invention.
A fourth aspect of the invention provides a method of pixel processing in a pixel processor, the method comprising: subtracting a baseline value from a pixel value to produce a calibrated pixel value; selectively adjusting a baseline reference value to produce said baseline value; and, in respect of a plurality of pixel values, adjusting said baseline reference value for some of said pixel values and not adjusting said baseline reference value for the rest of said pixel values.
In preferred embodiments, subtracting the integer value of the average of dark reference pixel values of a row from each image pixel value of that row results in slight row to row variation that gives rise to a row-based pattern that can be discernible to the human eye. In preferred embodiments of the invention, a further 1 is subtracted from the value of a percentage, or sub-set, of pixels in the respective row, where the percentage, or sub-set, is determined by the fractional part of the average value of dark reference pixels of the row. Advantageously, the percentage, or sub-set, of pixels are selected randomly, or pseudo-randomly, across the respective row. As a result, the undesired row-based pattern is eliminated, or substantially eliminated at least to the extent that the pattern is not discernible by the human eye. Advantageously, the resultant baseline for the image row is also more accurate as the average pixel baseline for the row has sub-integer precision. More generally, in preferred embodiments, an adjusted baseline reference value is used as the baseline value for a distributed sub-set of pixels in each row of the image sensor. The pixels of the sub-set are therefore calibrated differently to the other pixels in the row, and the distribution of the differently-calibrated pixels eliminates or reduces undesirable row-based patterns in images.
Preferred embodiments of the invention result in a smoother dark (background) image, which equates to higher image quality and a better user experience in comparison with images produced in conventional manner.
Brief Description of the Drawings
An embodiment of the invention is now described by way of example and with reference to the 5 accompanying drawings in which: Figure 1 is a schematic diagram of an image sensor system embodying one aspect of the invention, the system comprising a digital image sensor and a pixel processing device embodying another aspect of the invention; Figure 2 is an illustrated of a row of pixels of the image sensor of Figure 1 and Figure 3 is a schematic view of a preferred embodiment of the pixel processing device of Figure 1.
Detailed Description of the Drawings
Referring now to Figure 1 of the drawings there is shown, generally indicated as 10, an image sensor system embodying one aspect of the invention. The system 10 comprises an image sensor 12, a pixel processing device 14 embodying another aspect of the invention, and readout circuitry 16 for obtaining output data from the image sensor 12 and providing it to the pixel processing device 14. The pixel processing device 14 and readout circuitry 16 may be provided separately (e.g. implemented by a respective integrated circuit (IC)) or may be provided together (e.g. as part of the same IC). For example, the pixel processing device 14 and readout circuitry 16 may be implemented using the same field programmable gate array (FPGA). In some embodiments, the readout circuitry 16, or at least part of it, may be incorporated into the image sensor 12.
The image sensor 12 comprises an array 19 of pixels 18. In the illustrated embodiment the pixel array is a two-dimensional array comprising Y rows of pixels 18 and X columns of pixels 18. The values of X and Y can vary depending on the application. Each pixel 18 comprises a photodetector 30 or photosensor (not shown) that converts light into electrical charge or current.
The readout circuitry 16 is configured to obtain output data for each pixel 18 in the form of a pixel value that represents the level of light detected by the respective pixel 18 The image sensor 12 may be of any conventional type, for example it may be a CCD, CMOS or sCMOS image sensor. The type of the pixels 18 is also conventional and depends on the technology of the sensor 12 as would be apparent to a skilled person. The readout circuitry 16 may also be conventional and its configuration depends on the technology of the image sensor 12. For example, the readout circuitry 16 may comprise circuitry (not shown) connected to the image sensor 12 and configured to produce a digital pixel value for each pixel 18, typically depending on the level of electrical charge generated by the pixel 18. In some embodiments, the readout circuitry 16 may comprise analog-to-digital (AID) circuitry for converting analog pixel values provided by the sensor 12 into corresponding digital values. Alternatively: the image sensor 12 (e.g. in the case of an sCMOS sensor) may include ND circuitry for providing digitized pixel values as an output. The readout circuitry may include one or more readout register, and/or one or more multiplication register and/or one or more amplifier, and its specific composition and configuration can vary depending on the type of the image sensor 12. Depending on the technology of the sensor 12, all or part of the readout circuitry 16 may be included in the sensor 12. The readout circuitry 16 may comprise respective readout circuitry for each pixel 18, or for groups (e.g. rows) of pixels 18, or may comprise common readout circuitry for all pixels 18. For example, for a CMOS image sensor the circuitry 16 may comprise a respective amplifier for each pixel 18. In the illustrated embodiment: the circuitry 16 is assumed to comprise respective readout circuitry (not shown) for each row of pixels 18, each instance of readout circuitry comprising an amplifier. In preferred embodiments the image sensor 12 is assumed to be an sCMOS image sensor.
In preferred embodiments, the image sensor 12 includes at least one dark region and an image region IR. In the illustrated embodiment: the image sensor 12 includes first and second dark regions DR1, DR2. Each dark region DR1, DR2 comprises a plurality of the pixels 18, which may be referred to as dark reference pixels, held in a dark state (for example by covering or masking them such that they are not exposed to light incident on the sensor 12). The image region IR comprises a sub-array of the pixels 18, which may be referred to as image pixels: which are exposed to light in use and whose outputs are used to generate images. In preferred embodiments, the arrangement is such that each row of pixels 18 includes a plurality of dark reference pixels 18 and a plurality of image pixels 18. All rows of the array 19 preferably have the same arrangement of dark reference pixels and image pixels. Alternatively, the image sensor 12 may include one or more rows of dark reference pixels (that may include no image pixels).
In the illustrated embodiment, a first dark region DR1 is provided along a first side of the pixel array 19, extending parallel with the pixel columns such that a first end portion of each pixel row is located in the first dark region DR1. A second dark region DR2 is provided along a second side, opposite the first side, of the pixel array: extending parallel with the pixel columns such that a second end portion, opposite the first end portion, of each pixel row is located in the second dark region DR2. Figure 2 shows an exemplary row of pixels 18 from the pixel array 19 of Figure 1. Each row has a one dimensional array IF of image pixels 18 with a respective one dimensional array DP1, DP2 of dark reference pixels 18 at each end of the image pixel array IP. The number of image pixels and dark reference pixels in a row may vary from embodiment to embodiment. The size and location of the, or each: dark region may also vary from embodiment to embodiment.
The image pixels 18 of each row may be calibrated with respect to a baseline value. In preferred embodiments, the baseline value is determined using the values of the dark reference pixels of the 40 respective row. One method of calibrating the image pixels involves calculating an average value of the dark reference pixel values within a row of pixels 18, and subtracting the average value from
S
each of the image pixel values for the same row. In the present embodiment, the dark reference pixels 18 of each row go through the same row amplifier as the image pixels 18 for that row. Accordingly, once all the pixels 18 of all rows of the array 19 have been calibrated in this way, any row to row amplifier-related variation should be eliminated or substantially eliminated. giving an uniform row baseline across the entire image. However, it is found that subtracting the integer value of the average of dark reference pixel values in this way results in slight row to row variation in image pixel values that gives rise to a row based pattern that can be discernible to the human eye in the resulting image. This gives a user the perception that areas of darkness in the image are noisy and that the image quality is sub-optimum.
In preferred embodiments of the invention, and as is described in more detail with reference to Figure 3. a further 1 is subtracted from a percentage of the image pixel values of the respective row of pixels 18, where the percentage is determined by a fractional part of the average value of the dark reference pixel values of that row, and wherein the pixels 18 from whose pixel values the further 1 is subtracted are selected randomly or pseudo-randomly from the image pixels 18 of the respective row. This processing of the pixel values compensates for the effects of the hardware-based row to row variations and eliminates, or substantially eliminates, the undesirable row based pattern in the resulting image.
In preferred embodiments, processing of the pixel values 18 in order to perform baseline calibration and hardware-related compensation is performed by the pixel processing device 14. Typically, the pixel processing device 14 is implemented in electronic hardware, for example as one or more integrated circuit (IC). In preferred embodiments, the pixel processing device 14 comprises a suitably configured field programmable gate array (FPGA). Advantageously, a hardware implementation of the pixel processing device 14 allows its operation to match the processing speed of the image sensor and not to add latency to the system 10.
Figure 3 shows a preferred embodiment of the pixel processing device 14. The pixel processing device 14 comprises at least one instance of a pixel processor 15. Figure 3 illustrates one instance 15A of the preferred pixel processor 15. One or more additional instances 158, 15C of the pixel processor 15 may be provided in order to process multiple pixel values in parallel. For example, a respective instance of the pixel processor 15 may be provided for each row of the array 19 so that the pixel values of all of the rows can be processed in parallel. Alternatively, or in addition, a respective instance of the pixel processor 15 may be provided for each image pixel 18 of a row so that the image pixel values of any given row can be processed in parallel.
The pixel processor 15 has an input for receiving an image pixel value from the readout circuitry 16, and at least one input for receiving a dark reference value. In preferred embodiments the dark reference value is received from the readout circuitry 16. The dark reference value serves as a baseline reference value from which the baseline value used to calibrate image pixel values is determined. as is described in more detail hereinafter.
The pixel values are integer values, and may conveniently be provided in binary form. The dark reference value comprises an integer part 20 and a fractional part 22, and may conveniently be provided in binary form. The dark reference value is calculated from dark reference pixel values. In the present embodiment, the pixel processor 15 is configured to process pixel values on a row-byrow basis and so the dark reference value is calculated from the dark reference pixel values of the row being processed. In preferred embodiments, the dark reference value is an average of the dark reference pixel values of the respective row of pixels. In alternative embodiments, the dark reference value may be calculated or provided in any other convenient manner, preferably using some or all of the dark reference pixel values of the sensor, and preferable dark reference pixel values that are captured at the same time as the image that is being processed. For example, the dark reference value may be calculated as an average of some or all of the dark reference pixel values of the sensor, and the same dark reference value may be used by the pixel processor 15 to process all rows of the array. Alternatively, the dark reference value may be a fixed value, or a user selectable value, or a user-adjustable value provided to the processor 15 (for example this may be appropriate in embodiments in which the image sensor 12 does not have any dark regions). However, as described above, in preferred embodiments the dark reference value is calculated dynamically for each row using dark reference pixel values for that row since it this gives a more accurate value for darkness for the respective captured image.
The preferred pixel processor 15 processes image pixel values one at a time in conjunction with the relevant dark reference value. The pixel processor 15 may process each image pixel value of a row, one at a time, in conjunction with the dark reference value for that row, or multiple instances of the processor 15 may process a respective image pixel value of the row in parallel depending on the configuration of the pixel processing device 15. It will be seen that the preferred pixel processor 15 uses the integer part 20 and the fractional part 22 of the dark reference value separately, and so these parts 20, 22 may be provided to the processor 15 by separate inputs, or may be separated by the processor 15 as is convenient.
The pixel processor 15 uses the received dark reference value to provide a baseline value for calibrating the image pixel values. The preferred pixel processor 15 includes a subtractor 24 for subtracting the baseline value from each image pixel value. The subtractor 24 receives each image pixel value and subtracts from it the baseline value to produce a corresponding calibrated output image pixel value. The pixel processor 15 includes a baseline generator 26 for providing the subtractor 24 with a variable baseline value that is based on the dark reference value received by the processor 15. In preferred embodiments, the baseline generator 26 provides the subtractor 24 with either a non-adjusted baseline value or an adjusted baseline value. In preferred embodiments, the non-adjusted baseline value comprises the received dark reference value, preferably just the integer part of the received dark reference value. The adjusted baseline value may be obtained by adding 1 to the non-adjusted baseline value. In preferred embodiments, the adjusted baseline value is obtained by adding 1 to the integer part of the received dark reference value. To this end, the baseline generator 26 includes a value adjustor 28, which in the illustrated embodiment adds either 1 or 0 to the integer part of the received dark reference value to generate the baseline value for the subtractor 24.
The baseline generator 26 includes means for determining, for each image pixel value that is processed by the pixel processor 15, whether or not to provide the subtractor 24 with the non-adjusted baseline value or the adjusted baseline value. The aim is to use the adjusted baseline value with some pixel values but not others. In preferred embodiments, the adjusted baseline value is used with some pixel values of the respective row but not with others of the same row. Preferably, the adjusted baseline value is subtracted from a percentage of the image pixel values of the respective row of pixels 18, where the percentage is determined by the fractional part 22 of the dark reference value.
Preferably, determining which pixel values are calibrated using the adjusted baseline value is performed randomly, typically pseudo-randomly. To this end the preferred baseline generator 26 includes a random number generator (RNG) 30. The RNG 30 has a seed input and generates a random number, more particularly a pseudo-random number, as an output depending on the value of the seed input. The RNG 30 may be implemented using any conventional means, for example it may comprise a polynomial based psuedo random bit sequence (PRBS) generator. In order to ensure that the RNG 30 does not start at the same seed at the start of each row of pixels 18, which could introduce a perceptible column based pattern into the image, it is preferred that the fractional part of the received dark reference value is used to provide the seed input value. However, any other seed value may alternatively be used.
In some embodiments, for example where multiple instances of the processor 15 are provided to process pixel values in parallel, to ensure that each instance of the processor 15 has a different seed value, the bits of the fractional part of the receieved dark reference value may be shuffled. Any conventional bit shuffling device 34 may be provided for this purpose.
The baseline generator 26 includes a comparator 36 for comparing the output of the RNG 30 with a reference value, and depending on the result of the comparison, to cause the value adjustor 28 to adjust the baseline value or not, i.e. to add 1 or 0 to the integer part of the received dark reference value in the illustrated example. The comparison may be a simple determination as to whether or not the output of the RNG 30 is greater than the comparator reference value. For example, if the output of the RNG 30 is determined not to be greater than the comparator reference value. then 1 is added to the received dark reference value, otherwise 0 is added (i.e. no adjustment takes place). In preferred embodiments the reference value provided to the comparator 36 is the fractional part 22 of the received dark reference value. The RNG 30 is preferably configured to randomly generate a number between 0 and 1. The resulting comparisons of the outputs of the RNG 30 with the fractional part 22 of the dark reference value will cause a percentage of the image pixel values to be calibrated using the adjusted baseline value, where the percentage is determined by the fractional part 22 of the dark reference value: and will also cause said percentage of image pixel values to be randomly distributed in the row, i.e. the pixel values that are calibrated using the adjusted baseline value are randomly, or more particularly pseudo-randomly, distributed across the respective row of pixels 18, i.e. relate to pixels in the row that are randomly or pseudo-randomly interspersed with the other image pixels in the row. In alternative embodiments, the pixel values that are calibrated using the adjusted baseline value may be distributed in any other convenient manner, e.g. evenly distributed in the respective row.
Accordingly. in preferred embodiments, the fractional part 22 of the dark reference value determines the percentage of image pixels in respect of which the baseline value is adjusted (increased). The RNG 30 produces a number between 0 and 1 which is compared to the percentage value. For example, if the fractional part 22 is 0.75, then when the output of the RNG 30 is less than or equal to 0.75 (which will statistically be the case 75% of the time), the corresponding image pixel value is processed using the adjusted (increased) baseline value, i.e. 75% of the image pixel values are processed using the adjusted baseline value and the corresponding image pixels are randomly distributed in the row.
Advantageously, the average baseline value subtracted across the pixel row is derived from both the integer and fractional parts of the reference mean giving a more accurate and proportional baseline value. This is achieved by ensuring the percentage equivalent of adjusted baseline values vs non-adjusted baseline values are equal to the percentage equivalent of the fractional part of the baseline value. For example, if the calculated dark reference is 100.75 then 75% of the pixels in the row will have 101 subtracted from them and 25% will have 100 subtracted. Across the entire row the average subtracted value is 100.75. In contrast, if only an integer value for the dark reference value was used, then 100 or 101 would be subtracted from every pixel and so the average baseline value subtracted would also either be 100 or 101. The preferred method gives a higher level of accuracy since the average value for darkness across an entire row and therefore the entire image, after calibration, will be closer to zero.
In preferred embodiments: the non-adjusted baseline value or the adjusted baseline value, as applicable, is subtracted from each pixel value to give the resultant calibrated pixel value. The received image pixel values are integer values, the baseline value is an integer value and the resultant calibrated pixel values are integer values.
Image sensor systems embodying the invention are particularly suited for use in digital cameras, particularly but not exclusively digital cameras incorporated into, or for incorporation into, microscope systems.
The invention is not limited to the embodiment(s) described herein but can be amended or modified 40 without departing from the scope of the present invention.
Claims (17)
- CLAIMS: 1. A pixel processing device comprising at least one instance of a pixel processor, each instance of pixel processor comprising: a subtractor configured to subtract a baseline value from a pixel value to produce a calibrated pixel value; a baseline generator configured to provide said baseline value to said subtractor, said baseline generator comprising: an adjustor for selectively adjusting a baseline reference value to produce said baseline value; and means for determining whether or not to cause said adjustor to adjust said baseline reference value to produce said baseline value, wherein, in respect of a plurality of pixel values, said determining means is configured to cause said adjustor to adjust said baseline reference value for some of said pixel values and not to adjust said 15 baseline reference value for the rest of said pixel values.
- 2. The device of claim 1, wherein said determining means is configured to determine randomly, or pseudo-randomly, for which of said pixel values said baseline reference value is adjusted.
- 3. The device of claim 2, wherein said determining means comprises a random number generator that is operable to output a random number for each of said pixel values, and a comparator for comparing each random number with a comparator reference value, and wherein for each of said pixels, said determining means is configured to cause said adjustor to adjust said baseline reference value or not depending on the comparison of the respective random number with the comparator reference value for the respective pixel value.
- 4. The device of claim 3, wherein said random number generator has a seed input from which random numbers are generated, and wherein said seed input is provided by said comparator reference value.
- 5. The device of claim 4, wherein said comparator reference value comprises data bits, and wherein said baseline generator includes a bit shuffling device for shuffling the bits of said comparator reference value.
- 6. The device of any one of claims 1 to 5, wherein adjusting said baseline reference value comprises adding 1 to said baseline reference value.
- 7. The device of any preceding claim, wherein said baseline reference value comprises an integer part of a dark reference value.
- 8. The device of any one of claims 3 to 5 or of claims 6 or 7 when dependent on claim 3, wherein said comparator reference value comprises a fractional part of a dark reference value.
- 9. The device of any preceding claim, wherein said plurality of pixel values correspond to a row of pixels, and wherein said determining means is configured to cause the pixel values in respect of which the baseline reference value is adjusted to correspond with pixels in said row that are interspersed with other pixels in said row that correspond with pixel values in respect of which the baseline reference value is not adjusted.
- 10. The device of claim 9, when dependent on claim 2, wherein said determining means is configured to cause the pixel values in respect of which the baseline reference value is adjusted to correspond with pixels in said row that are randomly or pseudo-randomly interspersed with other pixels in said row that correspond with pixel values in respect of which the baseline reference value is not adjusted.
- 11. An image sensor system comprising an image sensor comprising an array of pixels, a pixel processing device as claimed in any one of claims 1 to 10, and readout circuitry for obtaining pixel values from said image sensor and providing said pixel values to said pixel processing device.
- 12. The system of claim 11, wherein said plurality of pixel values comprises image pixel values from a row of said array of pixels
- 13. The system of claim 11 or 12, wherein said array comprises a plurality of dark reference pixels, and wherein said baseline reference value is calculated from, or otherwise derived from, the values 25 of at least some of said dark reference pixels.
- 14. The system of claim 13 when dependent on clam 12, wherein each row of said array includes a plurality of dark reference pixels, and wherein said baseline reference value comprises an integer part of a dark reference value calculated from, or otherwise derived from, the dark reference pixel 30 values of the respective row.
- 15. The system of claim 14, wherein said dark reference value is an average of the dark reference pixel values of the respective row.16. The device of claim 1 or the system of 12, wherein the baseline reference value comprises a dark reference value. the dark reference value comprising either a fixed value, a user-adjustable value, or a user-selected value or a value calculated from one or more dark reference pixel values.
- 16. A digital camera comprising an image sensor system as claimed in any one of claims 11 to 15.
- 17. A method of pixel processing in a pixel processor, the method comprising: subtracting a baseline value from a pixel value to produce a calibrated pixel value; selectively adjusting a baseline reference value to produce said baseline value; and, in respect of a plurality of pixel values, adjusting said baseline reference value for some of said pixel values and not adjusting said baseline reference value for the rest of said pixel values.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB1917353.3A GB2589362B (en) | 2019-11-28 | 2019-11-28 | Method and apparatus for processing image sensor output to improve image quality |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB1917353.3A GB2589362B (en) | 2019-11-28 | 2019-11-28 | Method and apparatus for processing image sensor output to improve image quality |
Publications (3)
Publication Number | Publication Date |
---|---|
GB201917353D0 GB201917353D0 (en) | 2020-01-15 |
GB2589362A true GB2589362A (en) | 2021-06-02 |
GB2589362B GB2589362B (en) | 2022-10-26 |
Family
ID=69105902
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
GB1917353.3A Active GB2589362B (en) | 2019-11-28 | 2019-11-28 | Method and apparatus for processing image sensor output to improve image quality |
Country Status (1)
Country | Link |
---|---|
GB (1) | GB2589362B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7626623B2 (en) * | 2003-09-09 | 2009-12-01 | Canon Kabushiki Kaisha | Signal processing apparatus, signal processing method, program, and storage medium employing random number generation |
US20150312499A1 (en) * | 2014-04-29 | 2015-10-29 | Semiconductor Components Industries, Llc | Imaging systems and methods for mitigating pixel data quantization error |
-
2019
- 2019-11-28 GB GB1917353.3A patent/GB2589362B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7626623B2 (en) * | 2003-09-09 | 2009-12-01 | Canon Kabushiki Kaisha | Signal processing apparatus, signal processing method, program, and storage medium employing random number generation |
US20150312499A1 (en) * | 2014-04-29 | 2015-10-29 | Semiconductor Components Industries, Llc | Imaging systems and methods for mitigating pixel data quantization error |
Also Published As
Publication number | Publication date |
---|---|
GB201917353D0 (en) | 2020-01-15 |
GB2589362B (en) | 2022-10-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7830423B2 (en) | Method and apparatus providing imager noise reduction using parallel input arithmetic mean modules | |
EP1961208B1 (en) | Method and apparatus for setting black level in an imager using both optically black and tied pixels | |
US8089533B2 (en) | Fixed pattern noise removal circuit, fixed pattern noise removal method, program, and image pickup apparatus | |
US8275213B2 (en) | Apparatus and method for gain correction | |
JP4327928B2 (en) | Electronic camera device | |
US20090278963A1 (en) | Apparatus and method for column fixed pattern noise (FPN) correction | |
JP2007259486A (en) | System and method for adaptive sensitivity control at each pixel of digital imager | |
JP2012090051A (en) | Imaging device and imaging method | |
US9648265B2 (en) | Imaging systems and methods for mitigating pixel data quantization error | |
KR20080106042A (en) | Solid-state imaging device and pixel correction method | |
US10867371B2 (en) | Fixed pattern noise mitigation for a thermal imaging system | |
GB2425229A (en) | Generation and storage of column offsets for an image sensor | |
KR20080080224A (en) | Method and apparatus for gray value identification for white balance | |
GB2589362A (en) | Method and apparatus for processing image sensor output to improve image quality | |
WO2012015628A2 (en) | Method for reducing image artifacts produced by a cmos camera | |
JPH0385972A (en) | Method and apparatus for digital image pickup | |
JP6568368B2 (en) | Imaging apparatus, control method therefor, program, and storage medium | |
US11431924B2 (en) | Image capturing apparatus and control method thereof | |
Jähne | Release 4 of the EMVA 1288 standard: Adaption and extension to modern image sensors | |
US10484626B2 (en) | Gain adjustment unit, recording medium holding gain adjustment program, endoscope and endoscope apparatus | |
JP2009105966A (en) | Fixed pattern noise removal circuit, fixed pattern noise removal method, program, and imaging device | |
Wocial et al. | A method to achieve high dynamic range in a CMOS image sensor using interleaved row readout | |
JP2005094513A (en) | Original reader | |
US8477218B2 (en) | Method for correcting the space noise of an image sensor by luminescence limitation | |
Zin et al. | Nuclear Instruments and Methods in Physics Research Section A |