JP2008271316A - Image processing apparatus and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing technology for improving tone reproduction of a wider range of dark portion by simple operation processing. <P>SOLUTION: In the image processing method, an average value of pixel values in an image is calculated, and the maximum value and the minimum value of those pixel values are also calculated. Using a tone conversion characteristic of increasing the brightness of those pixels having pixel values falling in a range between the average value minus the maximum value and the minimum value than the other pixels outside the range, the pixel values in the image are replaced. The image processing apparatus includes a means for calculating the average value of the pixel values in the image, a means for calculating the maximum value of the pixel values in the image, a means for calculating the minimum value of the pixel values in the image, and a means of replacing the pixel values in the image by using the tone conversion characteristic of increasing the brightness of those pixels having pixel values falling in the range between the average value minus the maximum value and the minimum value than the other pixels outside the range. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image processing technique, and more particularly to improvement in reproducibility of a dark part.

  There is a need to improve the gradation reproducibility of dark areas of display images in digital televisions and personal computers. Conventionally, as a method for improving gradation reproducibility in a dark part, a method called Retinex is used in which a luminance signal is divided into an illumination component and a reflection component using a Gaussian filter. However, this Retinex method is excellent in expanding the gradation reproducibility in the dark part, but has a drawback in that it requires a Gaussian filter operation for each pixel and requires a large amount of arithmetic processing.

  Further, as a method with less arithmetic processing, a method of extending a dark portion gradation using a histogram or a method of expanding a dark portion gradation within a fixed range according to the area of the dark portion has been used. For example, the method described in Patent Document 1 obtains a gradation position frequency number distribution of a video signal, and performs pixel conversion in which the gradation position distribution is set wider at a location having a higher frequency number. .

However, in the method of reproducing dark part gradation in a fixed range according to the histogram and the area of the dark part, the gradation of the dark part can be expanded in a small range, but the gradation in the large part of the dark part cannot be expanded. There wasn't. This is because it was difficult to easily find a stable global process against the change in the shape of the distribution.
Japanese Patent Laid-Open No. 2002-84436 (page 5, FIG. 3)

  SUMMARY OF THE INVENTION An object of the present invention is to provide an image processing technique for improving gradation reproducibility in a wider dark area with simple arithmetic processing.

  In order to solve the above problems, an image processing method of the present invention calculates an average value of pixel values in an image, calculates a maximum value of pixel values in the image, and calculates a minimum value of pixel values in the image. And a gradation conversion characteristic that increases the luminance of pixels that take a pixel value in a section between a value obtained by subtracting the average value from the maximum value and the minimum value as compared with a pixel outside the section. Used to replace pixel values in the image.

  The image processing apparatus according to the present invention calculates a mean value of pixel values in the image, a means for calculating a maximum value of the pixel values in the image, and a minimum value of the pixel values in the image. And a gradation conversion characteristic that increases the luminance of a pixel having a pixel value in a section between a value obtained by subtracting the average value from the maximum value and the minimum value as compared with a pixel outside the section. And means for replacing pixel values in the image.

  According to the present invention, it is possible to obtain an image processing technique for improving the gradation reproducibility of a darker area in a wider range by simple arithmetic processing.

  Examples of the present invention will be described below.

A first embodiment of the present invention will be described with reference to FIGS.
FIG. 1 is a block diagram showing an embodiment of the present invention. This is an embodiment when applied to a personal computer. As an application block, a case where it is applied after an image quality adjustment circuit 1620 in a GPU which will be described later will be described.

  An output signal 1621 from the image quality adjustment circuit 1620 enters the YUV frame memory 10 and the Y signal pixel capturing circuit 20. The signal captured by the Y signal pixel capturing circuit 20 reaches the average value calculating circuit 30, the maximum and minimum value detecting circuit 40, the low gradation pixel number counting circuit 50, and the total pixel number counting circuit 60. The average value calculation result 31 calculated by the average value calculation circuit 30, the maximum value and minimum value detection result 41 calculated by the maximum value and minimum value calculation circuit 40, and the low gradation pixel count circuit 50 are calculated. The low gradation pixel count result 51 and the total pixel count result 61 calculated by the total pixel count circuit 60 reach the Y gradation conversion characteristic creation circuit 70, respectively. The Y gradation conversion characteristic creation circuit 70 creates a conversion curve characteristic from the average value, maximum value, and minimum value of the pixel values of the entire image. The Y gradation conversion instruction signal 71 generated by the Y gradation conversion characteristic generation circuit 70 reaches the UV gradation conversion characteristic generation circuit 90 and the pixel replacement processing circuit 100. The UV gradation conversion characteristic creating circuit 90 creates a conversion characteristic that changes the gradation of the UV signal in accordance with the change in the gradation of the Y signal. The UV gradation conversion instruction signal 91 created by the UV gradation conversion characteristic creating circuit 90 reaches the inter-pixel value processing circuit 100. In the inter-pixel value processing circuit 100, the pixel replacement process is performed on the output signal 11 from the YUV frame memory 10 by using the Y gradation conversion instruction signal 71 and the UV gradation conversion instruction signal 91.

  FIG. 2 shows an example of a curve characteristic created by the Y gradation conversion characteristic creation circuit 70 in FIG. In the example of FIG. 2, the curve characteristic has a convexity upward between the minimum value O point of pixel values and the (maximum value−average value) B point. An example of an upwardly convex curve characteristic is, for example, when the input pixel value is x and the output pixel value is y, the intermediate point between the minimum value and the first half of (maximum value−average value), that is, (minimum value + maximum value) -Average value) up to 2 points can be expressed by a function of y = x + √x (the square root of x, the same applies hereinafter), and the second half from the middle point is a vertical line obtained by subtracting the function from the middle point to y = x On the other hand, if it is connected at point A in FIG.

  When the equation is modified, (y−x) ^ 2 = x, where x is non-negative (^ 2 means square, and so on). This is in the category of conic curves. A function symmetric with y = x + √x with respect to a vertical line drawn on y = x from the middle point is obtained by uv coordinates rotated by + 45 ° with respect to the xy coordinates, and when inversely transformed, x = y−√ It can be derived that the inverse function is (L−y). If it is difficult to determine the value of y on the inverse function, consider a line parallel to y = x passing through the x and y coordinates at a line-symmetrical position on y = x + √x as an auxiliary line.

  The upward convex basic function is y = f (x), the upward convex section length ((maximum value−average value) −minimum value) is L, and the convexity by the count result of the number of low gradation pixels is Assuming that the coefficient that affects the magnitude of the amplitude is A, a function that considers the change in amplitude and the change in interval length (suppresses the amplitude as the interval length increases) for the function y = f (x). = A × f (x) ÷ L Here, if the number of low gradation pixels below g gradation is n and the total number of pixels is ALL, it can be expressed as A = k × (n ÷ g) ÷ (ALL ÷ Dmax). The amplitude of the function of Y = f (x) can be changed according to the number of pixels. In the latter part (n ÷ g) (ALL ÷ Dmax), the pixel number density in the low gradation region is normalized by the pixel number density in the entire region. Note that Dmax is the maximum value of the dynamic range (255 in 8-bit gradation representation), and k is a constant. For example, when k = αL is set, roughly, Y≈αf (x), and α has a property of increasing the convex in inverse proportion to L or suppressing the convex amount in proportion to L. . Instead of Dmax, a value close to Dmax, a maximum brightness value calculated in the screen, or a value close to the maximum brightness value calculated in the screen may be used. In addition, the relationship between the minimum value of the dynamic range (0 value in the 8-bit gradation expression) and the minimum value of the brightness calculated in the screen is omitted in FIG. Whatever suits the way of thinking should be used.

  Further, the UV gradation conversion characteristic 90 can change the UV signal in accordance with the change of the Y signal by giving the same characteristic as FIG. 2 in consideration of the UV offset (± 128 gradation display). . In this description, the YUV signal has been described, but it goes without saying that the same effect can be obtained even if the RGB signal is processed.

  Furthermore, in this description, the processing was described as the next stage of image quality adjustment in the GPU, but it goes without saying that the same effect can be obtained by processing using software including the CPU, North Bridge, and Main Memory. . For example, FIG. 3 shows the flow of a video signal of a personal computer as in the prior art.

  The configuration of FIG. 3 will be briefly described. The video signal output 1000 and the video signal input 1001 from the analog broadcast receiver are switched by the switch 1010 and digitized as a baseband signal (YUV) by the A / D & video decoder circuit 1100, and the South Bridge 1500 through the PCI bus 1101 to go into. South Bridge 1500 is connected to ODD1510 and HDD1520. The signal from the South Bridge 1500 is processed by the North Bridge 1400, CPU 1200, and Main Memory 1300, and reaches the GPU 1600. The signal entered into the GPU 1600 is adjusted in pixel shape by the square scaler circuit 1610, then the image quality balance is adjusted in the image quality adjustment circuit 1620, converted in signal by the YUV-> RGB conversion circuit 1630, and the α blend & scaler circuit 1640 After the image size is changed to the display size, dithering is performed after the data size is changed to 6 bits in the 8-> 6 bit reduction & dither circuit 1650. The image quality adjustment circuit 1620 performs general brightness adjustment and contrast adjustment.

  In the processing by this software, processing that takes into account that scaler and image quality adjustment processing will come later is performed as necessary. The effect of processing by software is that the GPU may be configured with a general-purpose chip.

  A second embodiment according to the present invention will be described with reference to FIG. 2 and FIGS. Description of the parts common to the first embodiment is omitted. FIG. 4 shows the flow of a video signal of a digital television.

  The configuration of FIG. 4 will be briefly described. The video signal output from the analog broadcast receiving unit 1000 and the video signal input 1001 are switched by the switch 1010, digitized as a baseband signal (YUV) by the A / D & video decoder circuit 1100, and sent to the back-end processor 2300. The digital broadcast signal received by the digital broadcast receiver 2100 is also sent to the back-end processor 2300 after the signal is demodulated by the MPEG2-TS decoder circuit 2200. The image size of each signal input to the back-end processor 2300 is adjusted by the scaler circuit 2310 and the image quality is adjusted by the image quality adjustment circuit 2320.

  Next, the tone conversion unit 2325 performs tone conversion using a curve. After that, it is converted to RGB signal by YUV-RGB conversion circuit 2330, then reduced by 2 bits by 10-> 8 conversion & gradation correction circuit 2340, and then FRC (frame rate control) is performed by gradation correction circuit . The tone-corrected signal is D / A converted by the D / A circuit 2400 and then sent to the display unit 2500. The image quality adjustment circuit 2320 generally includes a dark part gradation reproduction circuit in addition to contrast adjustment and brightness adjustment.

  FIG. 5 shows a configuration example when this embodiment is applied to a digital television. The output signal 2321 from the image quality adjustment circuit 2320 enters the YUV frame memory 10 and the Y signal pixel capturing circuit 20 of the gradation conversion unit 2325 surrounded by a broken line. FIG. 5 shows an example in which the present embodiment is applied to a digital television. The block configuration and algorithm are the same as those applied to the personal computer of FIG.

  Further, when a logarithmic function is used as f (x) −x instead of √x in the first embodiment, for example, if the curve OA in FIG. 2 is v = a * ln (u + 1) (ln is a natural logarithm), the curve AB Is a * ln (√2L−u + 1). If it is difficult to find a numerical solution of y when this is converted into xy coordinates, a numerical table is provided for each L by using a ROM (not shown) in the Y gradation conversion characteristic creating circuit 70 in advance. You should just pull.

  A third embodiment according to the present invention will be described with reference to FIGS. Description of parts common to the first and second embodiments is omitted. Typical conic curves mentioned in the first embodiment include an ellipse, a parabola, and a hyperbola, and an example in which each curve is used as f (x) is shown below.

  The formula of the ellipse is (u−L / √2) ^ 2 / a ^ 2 + (v + c) ^ 2 / b ^ 2 = 1 in uv coordinates (a, b, and c are positive constants). It can be set so that v = 0 when u = 0, √2L. In order to enter the first quadrant in the xy coordinates, a constant may be determined so that the slope of the tangent line at u = 0 is less than 45 °.

  The parabolic equation is v = −a (u−L / √2) ^ 2 + b in uv coordinates (a and b are positive constants). It can be set so that v = 0 when u = 0, √2L, and in order to enter the first quadrant with xy coordinates, a constant may be set so that the tangent slope at u = 0 is less than 45 °. .

  The hyperbolic equation is (x + a) (y−b) = − c (a, b and c are positive constants). x = 0, L and y = x can be set. FIG. 6 is a drawing example when L = 128. In the case of a = 16, b = 144, and c = 2304, the horizontal axis is x and the vertical axis is y.

  FIG. 7 shows an integer table of numerical values. The whole L = 32 (y32) when a = 4, b = 36, and c = 144 with respect to x, and some numerical values of L = 128 (y128) in FIG. 6 are represented as representatives. What is necessary is just to arrange about the function of appropriate a, b, and c according to each L.

In this example, f (x) is common and smooth on the vertical line.
In the above embodiment, the Y gradation conversion characteristic creation circuit creates the conversion curve characteristic using the average value, maximum value, minimum value, and amount of the low gradation region of the image, and performs pixel replacement processing based on the conversion curve characteristic. As a result, the tone reproducibility of the dark part was improved. As described above, it is possible to easily improve the gradation reproducibility of a dark part easily only by performing a simple arithmetic process similar to the conventional one. For example, in an image of a backlight scene, the maximum value of the pixel is not so low, but the average value is often considerably low. In such a case, it is possible to improve gradation reproducibility in a wide range of dark areas.

The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the invention. For example, a gradation conversion line may be used instead of the gradation conversion curve. The integer table can be said to be a piecewise gradation conversion straight line composed of about 256 line segments.

Various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above-described embodiments. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements according to different embodiments may be appropriately combined.

The block block diagram which shows one Example of this invention. An example of Y gradation conversion characteristics of the embodiment. PC video signal flow. The flow of the video signal of the digital television which shows 2nd Example of this invention. The block block diagram of the Example. The characteristic view which shows the 3rd Example of this invention. A schematic diagram of a numerical table used in the embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... YUV frame memory, 20 ... Y signal pixel acquisition circuit, 30 ... Average value calculation circuit, 40 ... Maximum value and minimum value detection circuit, 50 ... Low gradation pixel count circuit, 70 ... Y gradation conversion characteristic creation circuit , 90... UV gradation conversion characteristic creation circuit, 100... Pixel replacement processing circuit, 2325.

Claims (8)

Calculate the average pixel value in the image,
Calculating the maximum pixel value in the image;
Calculating the minimum pixel value in the image;
For the pixel that takes a pixel value in a section between the value obtained by subtracting the average value from the maximum value and the minimum value, the image is converted using a gradation conversion characteristic that increases the luminance as compared with a pixel outside the section. An image processing method characterized by replacing a pixel value in the image processing method. Calculate the average pixel value in the image,
Increases the luminance of pixels that take the pixel value in the interval between the maximum value that can be taken from the pixel value and the minimum value of the pixel value in the image, compared to the pixels outside the interval. An image processing method comprising replacing a pixel value in the image by using a gradation conversion characteristic to perform.   The image processing method according to claim 1, wherein the increase amount in the characteristic is monotonously decreased according to the length of the section. Calculating the number of pixels below the predetermined gradation in the image;
The image processing method according to claim 1, wherein the increase amount is monotonously increased according to a ratio of the number of pixels equal to or less than the predetermined gradation to the total number of pixels in the image.   The image processing method according to claim 1, wherein the conversion characteristic is a conic curve or a logarithmic function in the section.   6. The conversion characteristic according to claim 5, wherein the conversion characteristic is a conic curve in the section, and the conic curve is formed based on a value obtained by adding a square root value of the pixel value to the pixel value in the image. The image processing method as described. Means for calculating an average value of pixel values in the image;
Means for calculating a maximum value of pixel values in the image;
Means for calculating a minimum pixel value in the image;
For the pixel that takes a pixel value in a section between the value obtained by subtracting the average value from the maximum value and the minimum value, the image is converted using a gradation conversion characteristic that increases the luminance as compared with a pixel outside the section. An image processing apparatus comprising: means for replacing a pixel value in the image processing apparatus. Means for calculating an average value of pixel values in the image;
Means for calculating a maximum value of pixel values in the image;
Means for calculating a minimum pixel value in the image;
For the pixel that takes a pixel value in a section between the value obtained by subtracting the average value from the maximum value and the minimum value, the image is converted using a gradation conversion characteristic that increases the luminance as compared with a pixel outside the section. Means for replacing pixel values in
An image processing apparatus comprising: means for displaying an image of the replaced pixel value.

Images