CN101542582B - Method and apparatus for enhancing the dynamic range of an image - Google Patents

Abstract

A method of converting an image signal from an original dynamic range to a target dynamic range includes dividing the original dynamic range and the target dynamic range in to a plurality of corresponding sub-ranges and mapping each original sub-range to its corresponding target sub-range. One of the dynamic ranges is divided into a plurality of sub-ranges having equal size and the other dynamic range is divided into the corresponding sub-ranges each having a size based on a characteristic of the corresponding equal size sub-ranges. The invention may be employed in display, capture and image processing apparatus.

Description

Strengthen the method and apparatus of dynamic range of images

Technical field

The present invention relates to strengthen the method and apparatus of dynamic range of images, particularly a kind of method that strengthens the picture carrier dynamic range.

Background technology

Brightness is a metric that sends the luminous intensity or the quantity of light from the specific region, and brightness changes into the perceived brightness of scene or image more then.In real world, brightness range is in brightness and all be continuous on the time, and may be up to 14 orders of magnitude (14 powers of 10) from starlight to sunlight.Human eye can be watched the brightness range of 5 orders of magnitude.The image that produces high dynamic range (HDR) is readily appreciated that.For example, by adopting a plurality of time for exposure, photo can comprise the whole dynamic range of scene.The digital camera that can catch the high dynamic range scene becomes more and more welcome.But most display unit only can show the scene of 2 to 3 orders of magnitude.Brightness mapping (luminance mapping) is also referred to as tone mapping (tone mapping), can be used to dynamic range with real world be mapped to electronic display unit than low-dynamic range.Perhaps, but also have many algorithms and image technique to be developed to be used for dynamic range compression to an indication range with the HDR scene.An advantage of tone mapping is to optimize the data volume of video flowing in the pipeline.

The latest development of LCD display technology has produced and can show the display with high brightness dynamic image.But, since at image before showing on the display, a lot of images are switched to a lower dynamic range, thus need an inverse process that improves the luminance digital image dynamic range so that use with these high dynamic range displays.The direct method that enlarges dynamic range is only each pixel intensity value to be multiply by a constant.But this linear expansion is not considered characteristics of image and human visual system's characteristic.And the method that this linearity is amplified in proportion may produce noise, as producing contour effects (contouring effect) in the zone that gradually changes.

Also having a trend is exactly can watch video and image on portable electric appts such as mobile phone, personal digital assistant (PDAs) and game machine.Although expect that these equipment can display of high resolution and the image of high dynamic range, may determine the display device that uses in the portable set on resolution lower and illumination dynamic range, to move than standard display screen (as TV or computer display) as factors such as cost, design, restriction and battery lifes.

So, the purpose of this invention is to provide a kind of method and image processing equipment that strengthens the picture signal dynamic range.

Summary of the invention

In the present invention, a kind of picture signal is comprised from the method that initial dynamic range converts the target dynamic scope to: initial dynamic range and target dynamic scope are divided into the subrange of a plurality of correspondences, and each initial subrange is mapped to its corresponding target subrange.Be used at each initial subrange to the mapping function that shines upon between the corresponding target subrange or algorithm does not need and it is identical preferably not to be.

In order to obtain subrange, one of them dynamic range, that is to say or initial dynamic range or target dynamic scope, be divided into a plurality of subranges with equivalent size, and another dynamic range is divided into corresponding subrange, and the size of its each subrange is based on correspondent equal size subrange feature.

The present invention also comprises display, seizure and the image processing apparatus that adopts this method.

Only by the description of example, others of the present invention will be obvious all the more from following.

Description of drawings

Now by example also in conjunction with the accompanying drawings, exemplary of the present invention is described, wherein:

Fig. 1 describes a kind of first exemplary embodiments that initial and target dynamic scope is divided into the method for subrange before the tone mapping;

Fig. 2 describes the tone mapping of Fig. 1 subrange;

Fig. 3 describes a kind of second exemplary embodiments that initial and target dynamic scope is divided into the method for subrange before the tone mapping;

Fig. 4 is the module map of first exemplary embodiments of a high dynamic range display devices;

Fig. 5 is described in Fig. 4 display unit a low-dynamic range and target high dynamic range is divided into subrange, and

Fig. 6 is the module map of second exemplary embodiments of a high dynamic range display devices.

Embodiment

To describe exemplary embodiments of the present invention in detail now, and describe example in conjunction with the accompanying drawings.This method relates to a kind of method that strengthens the luminance dynamic range of image or picture signal.Image or picture signal can be still image or video image.This method is normally used when the bit-depth (bit-depth) of converted image or every pixel bit (bit-per-pixel), for example converts the 16-bit image to from the 8-bit image, and vice versa.But this is not a most critical for the purpose of the present invention, and this method can be used to strengthen the luminance dynamic range of image, and does not need to change the bit-depth of image.

According to the present invention, Fig. 1 describes a kind of method that strengthens the image brightness dynamic range to Fig. 3.The bit-depth of image is not considered, and may still remain unchanged or improve or reduce.Fig. 1 shows the statistic histogram of an image, along the pixel intensity dynamic range of horizontal x-axle at the image initial from minimum value (O-min) to maximum (O-max).It on vertical y-axle object brightness dynamic range from minimum value (T-min) to maximum (T-max).The 2nd y-axle remarked pixel quantity on histogram the right.Brightness is to send the luminous intensity of light or of quantity measures from special area, and brightness changes into the perceived brightness of image then, and it is by each pixel intensity (Luma) or gray level expressing.Nogata Figure 10 is illustrated in the pixel of some quantity in the picture signal and has brightness value on the x-axle.Brightness range on the x-axle is divided into 4 brightness subranges 1,2,3 and 4, and each brightness subrange has equal size (or size).The number of pixels that drops on the brightness value in each subrange is all different, and this can obviously find out from image histogram.The target dynamic scope is divided into corresponding 4 target subranges 5,6,7,8 on the y-axle.

The size of each target subrange 5,6,7,8 or size are dynamic, not necessarily equate.The size of each target subrange depends on the feature of initial subrange 1,2,3,4, and this feature is exactly to drop on the number of pixels of brightness in each initial subrange 1,2,3,4 in the example of present embodiment.For example, in Fig. 1, subrange 2 has the pixel of maximum quantity, thereby corresponding target subrange 6 has maximum size or size.Initial subrange 3 has the pixel of second most amounts, thereby corresponding target subrange 7 has the second largest size or the size of target subrange.Equally, target subrange 5 and 8 has a size or a size based on pixel quantity in the initial subrange 1 of correspondence and 4.In other words, the size of each target subrange is the proportional parts of a correspondence of whole target dynamic scope, corresponding to number of pixels in the corresponding initial subrange to total pixel number purpose percentage in the image.Therefore, according to this first exemplary embodiments, initial dynamic range is divided into the subrange of equivalent size, and the target dynamic scope dynamically is divided into the target subrange of a respective amount, and the size of each target subrange is directly proportional with the number of pixels with intensity level in the corresponding initial subrange.

With reference to Fig. 1, suppose that initial dynamic range is divided into 4 subranges 1,2,3,4 on an equal basis, and the number of pixels that hypothesis drops on the subrange is respectively N (1), N (2), N (3) and N (4), target subrange 5,6,7 and 8 size or size are respectively S (5), S (6), S (7) and S (8), and a kind of possible method that target subrange size is set is as follows:

$S\left(5\right)=\frac{N\left(1\right)(T_{\max }-T_{\min })}{N\left(1\right)+N\left(2\right)+N\left(3\right)+N\left(4\right)}$

$S\left(6\right)=\frac{N\left(2\right)(T_{\max }-T_{\min })}{N\left(1\right)+N\left(2\right)+N\left(3\right)+N\left(4\right)}$

$S\left(7\right)=\frac{N\left(3\right)(T_{\max }-T_{\min })}{N\left(1\right)+N\left(2\right)+N\left(3\right)+N\left(4\right)}$

$S\left(8\right)=\frac{N\left(4\right)(T_{\max }-T_{\min })}{N\left(1\right)+N\left(2\right)+N\left(3\right)+N\left(4\right)}$

With reference to Fig. 2, after the size of each initial subrange and corresponding target subrange is determined, a suitable tone mapping function T _k=F (O _k) be used to each the original intensity value O in the original intensity dynamic range _kBe mapped to a corresponding target brightness value T in the target dynamic scope _kThe mapping function F that is used for shining upon brightness value in every pair of subrange needs not be identical, may be linear or nonlinear.In the described embodiment of Fig. 2, Linear Mapping function is used for initial subrange 2 is mapped to target subrange 7, initial subrange 4 is mapped to 5, one Nonlinear Mapping functions of target subrange is used to initial subrange 1 is mapped to target subrange 5 and initial subrange 3 is mapped to target subrange 7.Should be noted that any known tone mapping function can be utilized to initial subrange is mapped to the target subrange in prior art.Disclose in co-inventor's previous patent application US 11/809,095 (in application on May 31st, 2007) lining and a kind of each initial subrange tone to be mapped to the method for its corresponding target subrange, be attached to herein by reference in this its whole contents.

Fig. 3 describes second exemplary embodiments of the present invention, and the target dynamic scope is divided into the subrange 11,12,13,14 of equivalent size on the wherein vertical y-axle.According to the special feature of image histogram, the size or the size of initial subrange 15,16,17,18 that can the Dynamic Selection correspondence, this special feature is again the number of pixels with intensity level in the initial subrange in this embodiment.In this example, select the size or the size of initial subrange 15,16,7,18, make that the number of pixels on each initial subrange is identical.Once more, in case determined the size of initial subrange 15,16,17,18 and target subrange 11,12,13,14, follow any known tone mapping method and can be used to pixel intensity is mapped to corresponding target subrange from initial subrange.

The hypothetical target subrange is divided into the subrange 15,16,17 and 18 of 4 equivalent size, the subrange of their correspondences in input dynamic range is respectively 11,12,13 and 14, and their size is respectively S (11), S (12), S (13) and S (14), the number of pixels that drops in 11,12,13 and 14 is respectively N (11), N (12), N (13) and N (14), and a kind of possible method that 11,12,13 and 14 sizes or size are set is to reach by minimizing following target function:

$\{S\left(11\right),S\left(12\right),S\left(13\right),S\left(14\right)\}=\arg \left(\min (\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{(S\left(i\right)-\frac{O_{\max }-O_{\min }}{4})}^2-\mathrm{\λ}\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{(N\left(i\right)-\frac{1}{4}\underset{j=1}{\overset{4}{\mathrm{\Σ}}}N\left(j\right))}^2)\right)$

Wherein 0＜λ＜1 is a constant value.A kind of similar by using by G Qiu, J.Guan, (ICRP 2006 in " Tone mapping for HDR image using optimization-A new closed from solution " literary composition for J.Duan and M.Chen, 18th International Conference on Pattern Recognition, 20-24August, 2006, Hong Kong) method of proposing can solve this optimization problem.

According to the present invention will be initial and the target dynamic scope be divided into subrange and respectively the tone benefit of shining upon each subrange be: the dynamic range of the subrange that the maximum number pixel drops on is extended, and the while, those had only the dynamic range of the subrange of small number of pixels to be compressed.Contrast and details on the image major part be can improve like this, and the order of magnitude or full luminance dynamic range do not needed to change.

The present invention can an applicable exemplary be that a standard picture will be displayed on the high dynamic range display device.The inventor has formerly applied for having proposed a kind of liquid crystal display with dynamic backlight in 11/707,517 (in the application on February 16th, 2007), and it can improve the contrast and the bit-depth of the luminance dynamic range that shows output.At this, be attached to herein by reference in the content of the described application 11/707,517 of on February 16th, 2007 application.In a preferred embodiment of the present invention, this liquid crystal display comprises an image brightness processor, be used for improving the dynamic range of low-light level dynamic range (LDR) image that receives, thereby image can by equipment with one more the form of high brightness dynamic range show, thereby improve visual contrast degree and details in the image.

At Fig. 4, show the module map of a high brightness dynamic range display device, usually with application 11/707,517 (in application on February 16th, 2007) in disclosure be the same.Display device has a variable intensity back light apparatus, is used to provide a liquid crystal display (LCD) panel of giving backlight.The LCD panel has a plurality of light to transmit display element.A lcd controller control light transmits the light transmission of display element.As known in the prior art of LCD display, lcd controller receives low-dynamic range (LDR) image of a standard, and correspondingly controls the light that each light transmits display element and transmit.The back light apparatus of LCD panel has a backlight, a plurality of light-emitting diodes and one is housed thereon is used for individually controlling the backlight controller of LEDs illumination.Equipment also comprises an image processor, is used for the LDR image transitions is become a high dynamic range (HDR) image, so that be input to backlight controller.Backlight controller receives the HDR image, and analyzes the HDR image and output signal to LEDs with generation, so that individually control LED brightness.By individually controlling the brightness of each LED, and in conjunction with the transmission of LCD element accordingly, can improve the visual luminance dynamic range of each element of display device, it is different from a backlight LCD display that tradition is constant, and image can be regarded a HDR image as.

Below discuss and to have described image processor and how to use and a kind ofly convert the LDR picture signal on the LCD display HDR picture signal according to method of the present invention.Fig. 5 shows an image histogram, and it has the pixel intensity dynamic range of the LCD picture signal from minimum value (I-min) to maximum (I-max) along horizontal x axle.The target HDR that the vertical y axle in edge is corresponding display brightness dynamic range from minimum value (D-min) to maximum (D-max).Display dynamic range is bigger than dynamic range of images.Nogata Figure 20 represents to have in the LDR picture signal pixel quantity or the number of brightness value on the x axle.Strength range along the x axle is divided into 4 intensity subranges 21,22,23 and 24, and each intensity subrange has identical size (or size).Can find out obviously that from image histogram the number of pixels that drops on intensity level in each subrange is different.Display dynamic range on the y axle is divided into 4 demonstration subranges 25,26,27,28 of corresponding number.Select to show the size of subrange 25,26,27,28, so that each subrange size is directly proportional with the number of pixels that drops on interior gray scale of correspondence image subrange or intensity level, as previously mentioned.After the size of each image subrange and corresponding demonstration subrange was determined, a suitable tone mapping function was used to each image subrange is mapped to separately the demonstration subrange of its correspondence.The tone mapping function that is used for shining upon every pair of subrange needs not be identical, and may be linear or nonlinear.Compare with previous example, the dynamic range of all subranges is increased in this example, and the result causes the integral body of dynamic range to increase.But this only is a feature of this special example, and reducing appears in the scope of a possible special subrange, although whole dynamic range is increased.

A prominent example of the present invention has below been described.But should be appreciated that conspicuous for a person skilled in the art modifications and variations should not be counted as exceeding scope of the present invention.A kind of such modification as shown in Figure 6.Should be noted that, be that the image of HDR form can be displayed on the equipment of previous application 11/707,517 (in application on February 16th, 2007) description.This equipment can comprise the tone mapping processor of a HDR to LDR, is used for input HDR image transitions is become the LDR form that is used by lcd controller and panel.In this example, the target dynamic scope will be littler than initial dynamic range.

Claims (6)

1. one kind converts picture signal the method for target dynamic scope to from initial dynamic range in image processing equipment, and this method comprises:

Initial dynamic range is divided into a plurality of initial subranges with equivalent size;

The target dynamic scope is divided into a plurality of target subranges, each initial subrange has the target subrange of a correspondence, the bigger target subrange of initial subrange correspondingly-sized that contained number of pixels is bigger, the less target subrange of initial subrange correspondingly-sized that contained number of pixels is less, and

Each initial subrange is mapped to its corresponding target subrange.

2. method according to claim 1, wherein initial dynamic range is the minimum brightness of image pixel and the scope between the high-high brightness, the target dynamic scope is the scope between minimum display brightness of image display and the maximum display brightness.

3. the display unit of a display image comprises:

A LCD panel has a plurality of light and transmits display element;

A lcd controller is used for controlling the light transmission that light transmits display element, to respond first picture signal with original intensity dynamic range;

A LED-backlit plate has a plurality of luminaires and is used for backlight illumination light and transmits display element;

A backlight controller according to second picture signal with object brightness dynamic range, is used for individually controlling the irradiation of luminaire;

An image processor able to programmely requires 1 described method with enforcement of rights, is used for picture signal that receives of conversion between original intensity dynamic range and object brightness dynamic range.

4. display unit according to claim 3, wherein initial dynamic range is the minimum brightness of image pixel and the scope between the high-high brightness, the target dynamic scope is the scope between minimum display brightness of image display and the maximum display brightness.

5. one kind is improved the luminance digital image dynamic range to improve the method for visual contrast degree and details in image in image processing equipment or image display, and this method comprises:

Initial dynamic range is divided into a plurality of initial subrange with equivalent size;

The target dynamic scope is divided into a plurality of target subranges, each initial subrange has the target subrange of a correspondence, the bigger target subrange of initial subrange correspondingly-sized that contained number of pixels is bigger, the less target subrange of initial subrange correspondingly-sized that contained number of pixels is less, and

Each initial subrange is mapped to its corresponding target subrange.

6. method according to claim 5, initial dynamic range is the minimum brightness of image pixel and the scope between the high-high brightness, the target dynamic scope is the scope between minimum display brightness of image display and the maximum display brightness.

Images