KR100783122B1 - Device and method for wide dynamic range correction - Google Patents

Abstract

An apparatus and a method for wide dynamic range correction are provided to adjust brightness distribution of an image evenly, thereby improving the visibility of an entire image, minimizing gradation loss which can be generated in a correction process, minimizing difference between blocks, and applying WDR(Wide Dynamic Range). An apparatus for wide dynamic range correction includes a block module, a lookup table generating module, a filter module and a correction module. The block module makes inputted images to blocks for classifying at least two areas. The lookup table generating module generates a block by a block lookup table with a correction output value for brightness. The filter module changes the generated block by a block lookup table to at least one lookup table and a correction lookup table. The correction module computes a correction output value of correction lookup tables, interpolates correction output values, and selects an output value of an associated pixel.

Description

Wide dynamic range correction device and correction method {DEVICE AND METHOD FOR WIDE DYNAMIC RANGE CORRECTION}

1 is a diagram schematically illustrating gamma correction, which is one of conventional contrast corrections.

2 is a diagram schematically showing a histogram equalization, which is another conventional contrast correction.

FIG. 3 is a diagram schematically illustrating an improved histogram equalization, another one of conventional contrast corrections.

4 is a block diagram showing a schematic configuration of an embodiment of a wide dynamic range correction apparatus of the present invention.

FIG. 5 is a diagram showing the blocking of an input image in one embodiment of the wide dynamic range correction apparatus of the present invention.

FIG. 6 is a diagram illustrating an example of generating a lookup table (LUT) of each block according to the block diagram shown in FIG. 5.

FIG. 7 shows another example in which an image is divided into blocks and a lookup table is formed for each block as an embodiment of the wide dynamic range correction device of the present invention.

FIG. 8 is a diagram illustrating each lookup table after filtering the lookup tables illustrated in FIG. 7 through a low pass filter.

FIG. 9 is a diagram schematically illustrating a method of interpolating a correction value from the lookup table shown in FIG. 8.

FIG. 10 is a view schematically showing a method of processing a histogram as another embodiment of the wide dynamic range correction apparatus of the present invention.

11 is a view showing a comparison of image quality improvement when the conventional method and the method of the present invention are applied.

The present invention relates to a wide dynamic range correction device and a correction method, and more particularly, in the case of an image in which light and dark areas are extremely distributed, such as backlighting, the light amount is not adjusted based on either of them. By correcting this, the contrast distribution can be made even in the case of such an image, thereby improving the overall visibility of the image, improving the identification of the light and dark portions, and minimizing the gradation loss that may occur in such correction. The present invention relates to a wide dynamic range correction apparatus and a correction method that can be provided.

Recently, as user demand for digital image processing apparatuses including digital cameras, digital camcorders, camera phones, etc. increases rapidly, their functions are increasingly diversified and intelligent.

However, when the image processing apparatus acquires an image in a state of backlight or the like, an image in which light and dark areas are extremely distributed, that is, an image in which a background is generally bright and a subject is dark is obtained on the screen. There was a problem that the image quality is degraded so that the subject is not properly identified.

Various methods have been proposed to solve this problem, and the methods of FIGS. 1 to 3 have been applied thereto.

That is, FIG. 1 schematically shows a backlight compensation through gamma correction. In this case, as shown in FIG. 1, there is a problem that a loss of bright data occurs and contrast of the list is eliminated, so that the highlight of this part is removed and the overall contrast is poor. There is a problem. In addition, the case shown in FIG. 2 schematically shows a backlight compensation through histogram equalization, through which a lookup table is formed and a correction is made using the same. As an example of this, there is a Korean Patent Application No. 1997-038701. In this case, when only one lookup table is applied to the entire image, and only the histogram equalization is applied, there is a problem in that overall grayscale loss occurs and a step portion of the contrast is not natural, and a stepped shape occurs. In this case, WDR is the most widely used method, and compresses the area of the existing histogram and applies the histogram equalization method to form a lookup table and corrects the image accordingly. Apply only one lookup table to the image As for the entire image, there is a problem that the gradation loss occurs in the arm and roll.

Therefore, in order to overcome such a problem, a method of applying a lookup table suitable for each area of an image may be proposed. In this case, image quality may be improved within each block, but the contrast may be increased at the boundary between the blocks. Since there is a problem that the boundary of the block is generated due to the contrast being different, it is necessary to improve this by applying a separate lookup table for each block of the image, while developing a correction device and a correction method that can obtain a natural image even at the boundary of the block. It is true.

In order to solve the problems of the prior art as described above, the present invention improves the overall visibility of the image with respect to the image in which the bright area and the dark area are extremely distributed, improves the identification of the ridge and the dark part, and minimizes the gray loss. An object of the present invention is to provide a wide dynamic range correction apparatus and a correction method capable of performing an improved wide dynamic range correction.

In order to achieve the above object, the present invention

In the wide dynamic range correction device for correcting the contrast of digitally captured images,

A blocking module for making an input image into a block for dividing the input image into at least two constant regions;

A lookup table generation module for obtaining a distribution of pixels of brightness of an image in each block divided by the blocking module and generating a lookup table for each block having a corrected output value for each brightness;

A filter module for converting the generated block-by-block lookup table into at least one lookup table adjacent to one side thereof and an averaged calibration lookup table; And,

The correction output values of the calibration lookup table of the block corresponding to the brightness value are obtained for each pixel of the block among the blocks, and the correction output values of the calibration lookup table of the at least one block adjacent to one side of the block are obtained. It provides a wide dynamic range correction device comprising a correction module for determining and outputting the output value of the pixel by interpolation.

In addition, the present invention

In the wide dynamic range correction method for correcting the contrast of a digital photographed image,

A blocking step of dividing an input image into at least two constant regions to make a block;

A lookup table generation step of obtaining a distribution of pixels for the brightness of the image in each block divided by the blocking step and generating a lookup table for each block having a correction output value for each brightness accordingly;

A filtering step of converting the generated block-by-block lookup table into at least one lookup table adjacent to one side thereof and an averaged calibration lookup table; And,

The correction output values of the calibration lookup table of the block corresponding to the brightness value are obtained for each pixel of the block among the blocks, and the correction output values of the correction lookup table of the at least one block adjacent to one side of the block are obtained. And a correction step of determining and outputting an output value of a corresponding pixel by interpolation.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wide dynamic range correction device, comprising: a block module for making an input image into a block for dividing an input image into at least two constant regions; A lookup table generation module for obtaining a distribution of pixels of the brightness of the image in each block divided by and generating a block-by-block lookup table having a corrected output value for each brightness, and adjacent to one side of the generated block-by-block lookup table. A filter module for converting at least one lookup table and an averaged lookup table, and a correction output value of a correction lookup table of a corresponding block corresponding to a brightness value for each pixel of the block among the blocks, Calibration lookup of at least one adjacent block Obtaining a corrected output value of the table each other interpolating the correction output value is configured to include a correction module for the output to determine the output value of the pixel.

Specific examples thereof are as shown in FIG. 4. That is, a block module for dividing an input image into a predetermined region in order to create a lookup table for each of them, rather than creating a lookup table for the entire input image, instead of dividing it into at least two blocks. Puts. Accordingly, as an example, an example divided into 16 blocks of 4 × 4 is shown in FIG. 5. For each block partitioned as above, the distribution of pixels for the brightness of the image in each block is obtained, and accordingly, a block-specific lookup table having a correction output value for each brightness is generated. As a method of generating such a lookup table, a lookup table generation method applied to various known wide dynamic ranges may be applied thereto.

Preferably, a known histogram equalization as shown in Fig. 2 and a lookup table forming method as shown in Fig. 3 can be applied thereto. An example of a look-up table (LUT) generated in this manner is illustrated in FIG. 6. More preferably, a value of a predetermined frequency or more in the distribution of pixels of the image brightness of the lookup table generation module is displayed only at the predetermined frequency, and the portion exceeding the predetermined frequency adds the cumulative value to the overall brightness distribution. The histogram can be modulated to generate a lookup table based on this method. Through this, the difference between the low frequency region and the high region can be reduced, and the slope of the frequency can be adjusted to induce a natural change of gray level in the block. Specific examples thereof are as shown in FIG. 10. That is, if the value of the pixel distribution for the brightness exceeds 380, it is cut and the frequency of the cut pixels is changed to a form of raising the floor level itself by 30 by averaging it over the entire brightness axis. This can prevent a sharp rise in the middle of the brightness axis in the cumulative distribution.

As shown in FIG. 7 (in this case, the block is not divided by 4 × 4), the LUT is clearly different in shape between different blocks in which contrast is contrasted. If LUT of this type is used as it is, contrast correction is well performed in the block, but the brightness of each block is clearly different, but the shape of the block is broken by block unit, which makes the whole screen unnatural. Therefore, in order to improve this, the filter module converts the generated lookup table for each block into at least one lookup table adjacent to one side thereof and an averaged calibration lookup table. That is, Fig. 8 shows the result of performing such filtering. The LUTs of FIG. 8 are LUTs one column below the LUTs of FIG. 7, for example, the LUT located below the left side of FIG. 8 is the LUT corresponding to itself before filtering and the LUTs of the upper, right and upper right thereof. It is obtained by averaging and acts as if passing through a kind of low pass filter (LPF), and the LUT located in the upper left of FIG. 8 becomes an LUT averaging four LUTs of FIG.

This is an example of averaging adjacent LUTs, besides simply averaging only with you and the LUTs on your right, or by averaging only you and your LUTs on your left and upper sides, or obtaining a calibrated LUT, Of course, you can also average the LUT and itself. Preferably, the adjacent blocks are three blocks of a) left or right side, b) top or bottom side, and c) diagonal side of the block, in terms of improvement of appropriate computation time and image quality, which are described in one direction. Of course, it must be made in the same direction continuously. A) left side, b) upper side and c) diagonal side or a) left side, b) lower side and c) diagonal side or a) right side, b) upper side and c) diagonal side or a The right side, b) lower side, and c) is one of the diagonal sides and one of these directions is then averaged to obtain the calibration LUT for each block.

For the correction LUT obtained as described above, the correction module obtains a correction output value of the correction lookup table of the corresponding block corresponding to the brightness value for each pixel of the block among the blocks, and corrects the lookup of at least one block adjacent to one side of the block. The correction output values of the table are obtained, and the correction output values are interpolated with each other to determine and output an output value of the corresponding pixel. That is, even when the difference is reduced through the filter module, when the calibration is performed only by the calibration LUT, the boundary between blocks may be left. In addition to the LUT, the LUT of the block adjacent to the block is utilized to obtain the correction output values of the pixels through the respective LUTs as described above, and interpolate them to use the correction output values of the actual pixels.

To do this, 1) simply obtain the correction output value of the pixel through the LUT of the block, and obtain another correction output value using the LUT of the left or right block of the block and consider it as the same distance (in this case, the simple average Or 2) obtain the correction output value of the pixel through the LUT of the block, and obtain the other correction output values using the LUTs of the two blocks on the left and right sides of the block, and consider these three values as the same distance. 3) obtain the correction output value of the pixel by interpolating the three values, or 3) obtain the correction output value of the pixel through the LUT of the block, and obtain other correction output values using the LUTs of the upper and lower blocks of the block. Consider these three values as the same distance and interpolate these three values to obtain the corrected output value of the pixel. 4) Through the LUT of the block The correction output value of the corresponding pixel can be obtained, and other correction output values can be obtained by using the LUTs of the upper, lower, left four blocks of the corresponding block, and these five values are regarded as the same distance to interpolate these five values to obtain the correction output value of the corresponding pixel. have. Preferably, as shown in FIG. 9, adjacent blocks may be a) a left or right side, b) an upper or a lower side, and c) a diagonal side corresponding to the quadrant in which the corresponding pixel exists from the center of the block. In this case, interpolation obtains the correction output value of the corresponding pixel through the LUT of the corresponding block, calculates the other correction output values by using the LUTs of three adjacent block blocks, and interpolates these four values by considering these four values as the same distance. The correction output value of the pixel pixel can also be obtained.

In this case, preferably, the interpolation of the correction module may be a weighted average method by adding a weight value that is contrary to the distance between the corresponding pixel and the center of each block to the correction output values obtained from each block lookup table. Various weighted average methods may be applied thereto, and the weight against the distance may be inversely proportional, but the present invention is not limited thereto and includes various forms that decrease with increasing distance. To this end, a separate distance-specific weight value table for weighting may be further included.

Preferably, in the case of interpolation using the adjacent block as shown in FIG. 9, as shown in FIG. 9, the output value through interpolation of the correction module is S1 * (1-y) + S2 * y (where S1 = A '* (1-x) + B' * x, S2 = C '* (1-x) + D' * x, where x is the pixel and the center of the block when the horizontal distance between blocks is 1 Is the relative horizontal distance of, y is the relative vertical distance between the pixel and the center of the block when the vertical distance between blocks is 1, A 'is the corresponding pixel correction output value by the lookup table in the block, B' is the block The corresponding pixel correction output value by the lookup table of blocks adjacent to the side from, C 'is the corresponding pixel correction output value by the lookup table of adjacent blocks from the block to the upper or lower surface, and D' is the lookup of adjacent blocks diagonally from the block. Corresponding pixel correction output value based on the table). In addition, although not shown in the drawings, in the case of calculating the vertical direction first and the horizontal direction later in a similar manner, the output value through interpolation of the correction module is S1 '* (1-x) + S2' * x (where S1 ' = A '* (1-y) + C' * y, S2 '= B' * (1-y) + D '* y).

In addition, the present invention provides a correction method through the correction device as described above, in the wide dynamic range correction method for correcting the contrast of the digital photographed image, by separating the input image into at least two constant areas A lookup table generation step of obtaining a block distribution of blocks and a distribution of pixels for the brightness of an image in each block divided by the blocking step, and generating a lookup table for each block having a correction output value for each brightness according to the generation; A filtering step of converting the prepared block-by-block lookup table into at least one lookup table adjacent to one side thereof and an averaged correction lookup table, and a correction lookup table of the corresponding block corresponding to a brightness value for each pixel of the block among the blocks. To get the correction output value, Even obtain a corrected output value of the one block of the correction look-up table to each other interpolating the corrected output values, including the correction step and outputting the determination of the output of the pixel constitutes a wide dynamic range correcting method.

Likewise, in the case of the correction method, as described above, a value of a predetermined frequency or more in the distribution of pixels with respect to the brightness of the image in the lookup table generation step is displayed only at the predetermined frequency, and a portion exceeding the predetermined frequency is stored in the accumulated value. The pixel distribution may be configured by averaging the overall brightness distribution, and the interpolation of the correction step is weighted and averaged by adding a weight that is opposite to the distance between the pixel and the center of each block to the correction output values obtained from each block lookup table. In the filtering step, the adjacent block is preferably a) left or right side, b) upper or lower side and c) diagonal side of the block, and in the correction step, the adjacent block is the center of the block. From a) left or right, b) up or down, depending on the quadrant from It is preferable that the side and c) are diagonal sides, and in this case, the output value through interpolation of the correction step is i) S1 * (1-y) + S2 * y (where S1 = A '* (1-x) + B '* x, S2 = C' * (1-x) + D '* x), or ii) S1' * (1-x) + S2 '* x, where S1' = A '* (1-y ) + C '* y, S2' = B '* (1-y) + D' * y), where x is the relative horizontal distance from the pixel when the horizontal distance between blocks is 1 and the center of the block , y is the relative vertical distance between the pixel and the center of the block when the vertical distance between blocks is 1, A 'is the pixel-corrected output value by the look-up table in the block, and B' is laterally adjacent to the block. The corresponding pixel correction output value by the lookup table of the block, C 'is the corresponding pixel correction output value by the lookup table of the block adjacent to the upper or lower surface from the block, and D' is the corresponding pixel correction output value by the lookup table of the adjacent block diagonally from the block. Pixel correction output value .

The comparison between the case where the correction apparatus and the correction method of the present invention is applied and the case where the conventional method is applied are shown in FIG. 11. The photo on the upper left side of FIG. 11 is according to the present invention. The lower left side shows gamma correction, the upper right side shows the method of FIG. 3, and the lower right side shows correction through histogram equalization of FIG. In this case, as shown in FIG. 11, in the case of gamma correction, the gray level is totally lost, and in particular, the contrast of the roster can be seen to be markedly reduced. In the case of histogram equalization, a step phenomenon occurs, The gradation loss of the dark part is large, and the improvement of the degree of gradation loss and the dark part of the upper part and the dark part is also performed in the case of the photograph of the upper right side.

According to the wide dynamic range correction device and the correction method of the present invention, in the case of an image in which light and dark areas are extremely distributed, such as backlight, the light is corrected for the entire range without adjusting the amount of light based on either one. Even in the case of burns, the roster is in the roster, and the dark portion is made of the dark portion, so that the overall visibility of the image is improved by improving the overall visibility of the roster and the roster, minimizing the loss of gradation that may occur in such correction. In addition, there is an effect that can provide a more improved image by minimizing the distinction between blocks according to the block, it can be applied to the WDR (Wide Dynamic Range).

The present invention described above is not limited to the above-described embodiment and the accompanying drawings, and various modifications of those skilled in the art without departing from the spirit and scope of the present invention described in the claims below. And modifications are also included within the scope of the invention.

Claims (10)

In the wide dynamic range correction device for correcting the contrast of digitally captured images, A blocking module for making an input image into a block for dividing the input image into at least two constant regions; A lookup table generation module for obtaining a distribution of pixels of brightness of an image in each block divided by the blocking module and generating a lookup table for each block having a corrected output value for each brightness; A filter module for converting the generated block-by-block lookup table into at least one lookup table adjacent to one side thereof and an averaged calibration lookup table; And, The correction output values of the calibration lookup table of the block corresponding to the brightness value are obtained for each pixel of the block among the blocks, and the correction output values of the calibration lookup table of the at least one block adjacent to one side of the block are obtained. And a correction module for determining and outputting an output value of the corresponding pixel by interpolation. The method of claim 1, The interpolation of the correction module is a weighted average of the correction output values obtained from the lookup table for each block by weighting the weight against the distance between the corresponding pixel and the center of each block weighted average. The method of claim 1, In the filter module, Adjacent blocks are a) left or right side of the block, b) top or bottom side and c) diagonal side, In the correction module, And the adjacent block is a) left or right side, b) upper or lower side, and c) diagonal side corresponding to the quadrant of the corresponding pixel from the center of the block. The method of claim 3, The output value through interpolation of the correction module is i) S1 * (1-y) + S2 * y, where S1 = A '* (1-x) + B' * x, S2 = C '* (1-x) + D '* x, or ii) S1' * (1-x) + S2 '* x, where S1' = A '* (1-y) + C' * y, S2 '= B' * (1 -y) + D '* y, where x is the relative horizontal distance from the pixel when the horizontal distance between the blocks is 1 and the center of the block, and y is the pixel and the corresponding block when the vertical distance between the blocks is 1 The relative vertical distance to the center of, A 'is the pixel correction output value by the look-up table in the block, B' is the pixel correction output value by the look-up table of adjacent blocks from the block, and C 'is And a corresponding pixel correction output value using a lookup table of blocks adjacent to the top or bottom surface, and D 'is a corresponding pixel correction output value from a lookup table of blocks adjacent to the diagonal side from the block. The method according to any one of claims 1 to 4, In the distribution of pixels for the brightness of the image of the look-up table generation module, a value of a predetermined frequency or more is displayed only at the predetermined frequency, and a portion exceeding the predetermined frequency is added to the total brightness distribution by averaging the accumulated value. Dynamic range correction device. In the wide dynamic range correction method for correcting the contrast of a digital photographed image, A blocking step of dividing an input image into at least two constant regions to make a block; A lookup table generation step of obtaining a distribution of pixels for the brightness of the image in each block divided by the blocking step and generating a lookup table for each block having a correction output value for each brightness accordingly; A filtering step of converting the generated block-by-block lookup table into at least one lookup table adjacent to one side thereof and an averaged calibration lookup table; And, The correction output values of the calibration lookup table of the block corresponding to the brightness value are obtained for each pixel of the block among the blocks, and the correction output values of the correction lookup table of the at least one block adjacent to one side of the block are obtained. And a correction step of determining and outputting an output value of the corresponding pixel by interpolation. The method of claim 6, The interpolation of the correction step is a weighted average of the correction output values obtained from the lookup table for each block by weighting the weight against the distance between the corresponding pixel and the center of each block. The method of claim 6, In the filtering step, Adjacent blocks are a) left or right side of the block, b) top or bottom side and c) diagonal side, In the correction step, And adjacent blocks are a) left or right side, b) upper or lower side, and c) diagonal sides corresponding to the quadrants of the corresponding pixels from the center of the block. The method of claim 8, The output value through interpolation of the correction step is i) S1 * (1-y) + S2 * y, where S1 = A '* (1-x) + B' * x, S2 = C '* (1-x) + D '* x, or ii) S1' * (1-x) + S2 '* x, where S1' = A '* (1-y) + C' * y, S2 '= B' * (1 -y) + D '* y, where x is the relative horizontal distance from the pixel when the horizontal distance between the blocks is 1 and the center of the block, and y is the pixel and the corresponding block when the vertical distance between the blocks is 1 The relative vertical distance to the center of, A 'is the pixel correction output value by the look-up table in the block, B' is the pixel correction output value by the look-up table of adjacent blocks from the block, and C 'is And a corresponding pixel correction output value using a lookup table of blocks adjacent to an upper surface or a lower surface, and D 'is a corresponding pixel correction output value using a lookup table of blocks adjacent to a diagonal side from the block. The method according to any one of claims 6 to 9, In the distribution of pixels for the brightness of the image in the lookup table generation step, a value of a predetermined frequency or more is displayed only at the predetermined frequency, and a portion exceeding the predetermined frequency is added to the total brightness distribution by averaging the accumulated value. Dynamic range correction method.

Images