JP2014109978A - Image processing device and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To increase the speed of histogram generation processing for generating a histogram data of a filter region set adjacent to a position of interest on an image.SOLUTION: A local histogram generation section that sequentially generates histogram data at each position on an image while shifting the position of interest one by one includes: a histogram storage 46 that stores histogram data at each position on the image; primary difference acquisition sections 51, 53 and 56 each of which acquires as a target primary difference data of a primary difference region where a filtering region at a position of interest and a filtering region preceding by one position are not overlapped with each other; and integration sections 52, 54 and 57 each of which integrates the primary difference data acquired by the primary difference acquisition section and the histogram data at a position preceding by one stored in the histogram storage to thereby acquire a histogram data at the position of interest.

Description

  The present invention relates to an image processing apparatus and an image processing method for generating histogram data indicating the frequency for each gradation and correcting an image based on the histogram data.

  A bright area and a dark area are mixed in the image captured by the imaging device.For example, a bright outdoor area and a dark indoor area due to solar radiation are included. Although a sufficient difference in brightness can be obtained between the two, there is a problem that the contrast is lowered in each region and the subject becomes unclear.

  To solve this problem, generate histogram data that shows the tone distribution of pixels that exist within a local area centered on a point of interest on the image, and emphasize the high-frequency tone contrast in the histogram data. There is known a tone correction technique based on a so-called local histogram equalization method, in which a tone conversion curve is obtained and tone correction is performed using the tone conversion curve. Thereby, the contrast is enhanced in each of the bright region and the dark region, and a clear image can be obtained.

  However, it is not practical to perform gradation correction by this local histogram equalization method on a pixel basis for the entire image because a huge amount of calculation is required. Therefore, the image is divided into a plurality of blocks, the representative values of each block included in the local area centered on each block are aggregated, histogram data of each block is generated, and gradation conversion is performed based on this histogram data. A technique for correcting a gradation by obtaining a curve is known (see Patent Document 1). Thereby, the amount of calculation can be reduced.

JP 2008-92052 A

  However, in the conventional technique for generating the histogram data in units of the blocks, it is necessary to repeat the process of generating the histogram data by adding up the representative values of all the blocks included in the filter area. There is a problem that the amount cannot be reduced sufficiently, and the gradation correction processing cannot be sufficiently speeded up. In addition, if the size of the block is increased, the amount of calculation can be reduced. However, in this case, the discontinuity at the boundary of the block becomes remarkable and the image quality is deteriorated. It is desirable to increase the speed.

  The present invention has been devised to solve such problems of the prior art, and the main object of the present invention is an image processing apparatus configured to increase the speed of histogram generation processing, and It is to provide an image processing method.

  An image processing apparatus according to the present invention generates histogram data for a filter region set around a position of interest on an input image, and corrects the image based on the histogram data A local histogram generator for sequentially generating the histogram data at each position while shifting the target position one by one on the image, and the local histogram generator at each position on the image The primary difference data is acquired for a primary difference area in which the histogram storage unit for storing the histogram data and the filter area at the position of interest and the filter area at the previous position do not overlap with each other. A primary difference acquisition unit, the primary difference data acquired by the primary difference acquisition unit, and one stored in the histogram storage unit The integrated histogram data at the position of a structure having, an integrated unit for acquiring histogram data at the target position.

  Further, the image processing method of the present invention generates an image of histogram data for a filter region set around a position of interest on the input image, and corrects the image based on the histogram data. A method of processing, comprising: sequentially generating the histogram data at each position while shifting the position of interest on the image one by one, wherein the step of generating the histogram data includes the filter region at the position of interest. Obtaining primary difference data for a primary difference area that does not overlap with the filter area at the previous position, and the primary difference data and histogram data at the previous position. And acquiring histogram data at the target position.

  According to the present invention, it is not necessary to add up the feature values (for example, luminance) at all positions included in the filter region in order to obtain the histogram data, so that the calculation burden can be reduced. The generation process can be speeded up.

Schematic configuration diagram showing an imaging apparatus according to the present embodiment Explanatory drawing explaining the condition of the gradation correction process performed in the gradation correction | amendment part 7 using the example of an actual image Flow chart showing a procedure of processing performed in the gradation correction unit 7 Explanatory drawing explaining the block division process performed in the block division part 10 Explanatory drawing explaining the local histogram generation process performed in the local histogram generation part 14 Explanatory drawing explaining the gradation conversion curve generation process performed in the gradation conversion curve generation part 16 Explanatory drawing explaining the gradation conversion process performed in the gradation conversion part 17, and the linear interpolation process performed in the linear interpolation part 18 The block diagram which shows schematic structure of the local histogram generation part 14 Explanatory drawing explaining a weighting area | region and a weighting coefficient Explanatory drawing explaining the point which produces | generates a weighted histogram Flow chart showing a procedure of local histogram generation processing performed in the local histogram generation unit 14 The block diagram which shows schematic structure of the weightless histogram generation part 31 Explanatory drawing explaining the outline of histogram generation processing without weight Flow chart showing the procedure of unweighted histogram generation processing Explanatory drawing explaining the outline | summary of a 1st area | region histogram production | generation process. Flow chart showing the procedure of first region histogram generation processing Explanatory drawing explaining the outline | summary of a 2nd area | region histogram production | generation process. Flow chart showing procedure of second region histogram generation processing Explanatory drawing explaining the outline | summary of a 3rd area | region histogram production | generation process Flow chart showing the procedure of third region histogram generation processing Explanatory drawing explaining the outline | summary of a 4th area | region histogram production | generation process. Explanatory drawing explaining the outline | summary of a 4th area | region histogram production | generation process. Explanatory drawing explaining the outline | summary of a 4th area | region histogram production | generation process. Explanatory drawing explaining the 4th field histogram generation processing in detail Flow chart showing procedure of fourth region histogram generation processing

  According to a first aspect of the present invention, there is provided a first invention for generating histogram data for a filter region set around an attention position on an input image, and generating the image based on the histogram data. An image processing apparatus that includes a local histogram generation unit that sequentially generates the histogram data at each position while shifting the target position one by one on the image, and the local histogram generation unit includes: Targeting a histogram storage unit that stores the histogram data at each position on the image, and a primary difference area in which the filter area at the target position and the filter area at the previous position do not overlap. A primary difference acquisition unit that acquires the primary difference data, the primary difference data acquired by the primary difference acquisition unit, and the histogram By integrating the histogram data at the previous locations stored in the accommodated section, a structure having, an integrated unit for acquiring histogram data at the target position.

  According to this, since it is not necessary to add up the feature values (for example, luminance) at all positions included in the filter area in order to obtain the histogram data, it is possible to reduce the calculation burden, and thereby the histogram generation process Can be speeded up.

  Here, the histogram data indicates the frequency for each feature value obtained by aggregating the feature values at each position in the filter area, and the primary difference data means each position in the primary difference area. This indicates the frequency for each feature value obtained by aggregating the feature values.

  The primary difference area includes a decrease amount deviating from the filter area at the target position in the filter area at the previous position, and an increase amount newly added to the filter area at the target position. When integrating data and histogram data, the reduced primary difference data is subtracted from the histogram data, and the increased primary difference data is added to the histogram data.

  In the second invention, the local histogram generation unit performs processing for one line for sequentially generating the histogram data at each position while shifting the position of interest in the first direction, orthogonal to the first direction. The primary difference storage unit stores the primary difference data acquired by the primary difference acquisition unit, and the primary difference data at the position of interest. A secondary difference acquisition unit that acquires secondary difference data for a secondary difference area that is not overlapped with the primary difference data at the position immediately before in the second direction; The primary difference acquisition unit includes the secondary difference data acquired by the secondary difference acquisition unit, and the primary difference data at a position immediately before in the second direction stored in the primary difference storage unit. To integrate the first order difference at the position of interest A structure in which to get over data.

  According to this, in order to acquire the histogram data, it is not necessary to add up the feature values of all the positions included in the filter area, and in addition, all the elements included in the primary difference area to acquire the primary difference data. Since it is not necessary to add up the feature values at the positions of the positions, it is possible to further reduce the calculation burden, thereby further speeding up the histogram generation process.

  Here, the secondary difference data indicates the frequency for each feature value obtained by aggregating the feature values at each position in the secondary difference area.

  In the secondary difference area, the decrease in the primary difference area at the position immediately before in the second direction deviates from the primary difference area at the target position, and the primary difference area at the target position is newly added. When the secondary difference data and the primary difference data are integrated, the reduced secondary difference data is subtracted from the primary difference data, and the increased secondary difference data is 1 Added to the next difference data.

  According to a third aspect of the present invention, the local histogram generation unit includes a first region histogram generation unit for a start position located at a corner of the image, and a first region histogram extending in the first direction from the start position. A second region histogram generator for a region excluding the start position in the line, and a third region histogram generation for a region excluding the start position in one line extending from the start position in the second direction. And a fourth region histogram generation unit for a region excluding a target region of the first region histogram generation unit, the second region histogram generation unit, and the third region histogram generation unit, The one-region histogram generating unit obtains the histogram data by collecting feature values at each position in the filter region at the start position, The region histogram generation unit includes the primary difference acquisition unit and the integration unit, and shifts the target position in the first direction from a position shifted by one in the first direction with respect to the start position. Histogram data is sequentially generated, and the third region histogram generation unit includes the primary difference acquisition unit and the integration unit, and takes attention from a position shifted by one in the second direction with respect to the start position. The histogram data is sequentially generated while shifting the position in the second direction, and the fourth region histogram generation unit includes the secondary difference acquisition unit, the primary difference acquisition unit, and the integration unit, and the start A process for one line for sequentially generating the histogram data at each position while shifting the position of interest in the first direction from a position shifted one by one in the first direction and the second direction with respect to the position, 1 in the second direction A configuration that is repeated while shifting by Inn.

  This makes it possible to efficiently generate histogram data at all positions in the image.

  According to a fourth aspect of the present invention, gradation correction is performed on the image, and the image correction unit includes a gradation conversion curve generation unit that generates a gradation conversion curve based on the histogram data; And a gradation conversion unit that performs gradation conversion on each pixel using a gradation conversion curve.

  According to this, when a bright region and a dark region are mixed in the image, it is possible to obtain a clear image by enhancing the contrast in each region.

  The fifth invention further comprises a block dividing unit that divides the image into a plurality of blocks, and a block representative value obtaining unit that obtains a representative value for each block from the values of the pixels in the block, The local histogram generator is configured to generate histogram data for each block.

  According to this, since the histogram generation processing is performed in units of blocks, it is possible to reduce the amount of calculation and increase the processing speed.

  According to a sixth aspect of the present invention, the local histogram generator generates weighted histogram data weighted according to the distance from the target position, and has different sizes with different weighting factors set. An unweighted histogram generator for generating unweighted histogram data for a plurality of filter regions, and the unweighted histogram data for each of the plurality of filter regions are integrated by multiplying each by the weight coefficient. And a weighted histogram generator for generating weighted histogram data.

  According to this, it is possible to adjust the influence of the feature value of each position on the histogram data according to the distance from the target position.

  According to a seventh aspect of the present invention, there is provided an image processing method for generating histogram data for a filter region set around a target position on an input image and correcting the image based on the histogram data. A step of sequentially generating the histogram data at each position while shifting the target position one by one on the image, and the step of generating the histogram data includes the filter region at the target position and the filter region The step of acquiring primary difference data for a primary difference area that does not overlap the filter area at the previous position, and the primary difference data and the histogram data at the previous position are integrated. And acquiring histogram data at the target position.

  According to this, similarly to the first invention, it is possible to speed up the histogram generation processing.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a schematic configuration diagram illustrating an imaging apparatus according to the present embodiment. This imaging device includes a lens 1, a diaphragm unit 2, an imaging device 3, an A / D converter 4, a preprocessing unit 5, a YC separation unit 6, and a gradation correction unit (image processing device) 7. , Y signal processing unit 8 and C signal processing unit 9.

  The lens 1 focuses light from the subject on the image sensor 3. The diaphragm unit 2 controls the amount of light incident on the image sensor 3. The image sensor 3 performs photoelectric conversion on the formed image. The A / D converter 4 converts an analog signal obtained from the image sensor 3 into a digital signal. The pre-processing unit 5 performs processing such as OB subtraction and white balance adjustment on the A / D converted image signal. The YC separation unit 6 separates the preprocessed image signal into a luminance signal and a color difference signal. The gradation correction unit 7 performs gradation correction on the luminance signal and the color difference signal output from the YC separation unit 6. The Y signal processing unit 8 performs necessary signal processing on the luminance signal. The C signal processing unit 9 performs necessary signal processing on the color difference signal.

  In the present embodiment, the imaging apparatus includes the gradation correction unit 7. However, the gradation correction unit is configured by an image processing apparatus different from the imaging apparatus, and an appropriate storage medium or communication medium is provided. The imaging data may be sent from the imaging device to the image processing device. In this case, the image processing apparatus can also be realized by introducing a gradation correction application into an information processing apparatus such as a PC.

  Next, the tone correction process performed by the tone correction unit 7 will be described. FIG. 2 is an explanatory diagram for explaining the state of gradation correction processing performed by the gradation correction unit 7 shown in FIG. 1 using an example of an actual image.

  In this image, a bright area and a dark area are mixed, that is, an area where a bright outdoor area is reflected by solar radiation and an area where a dark indoor area is captured are included. As shown in FIG. In the input image before correction, a sufficient contrast between the bright area and the dark area is obtained, but the contrast is lowered in each area, and the subject is unclear. When gradation correction processing is performed on such an input image by the gradation correction unit 7, as shown in FIG. 2B, the contrast is enhanced in each of the bright area and the dark area, so that the image is clear. Can be obtained.

  As shown in FIG. 1, the gradation correction unit 7 includes a block dividing unit 10, a block representative value calculating unit 11, a gradation number reducing unit 12, a block representative value storing unit 13, and a local histogram generating unit. 14, a histogram shaping unit 15, a gradation conversion curve generation unit 16, a gradation conversion unit 17, a linear interpolation unit 18, a ratio calculation unit 19, and multipliers 21 and 22. The gradation correction unit 7 receives image data output from the preprocessing unit 5.

  Hereinafter, processing performed in each unit of the gradation correction unit 7 will be described. First, a procedure of processing performed in each unit of the gradation correction unit 7 will be described. FIG. 3 is a flowchart showing a procedure of processing performed by the gradation correction unit 7.

  In the gradation correction unit 7, first, a process of dividing the input image into a plurality of blocks is performed in the block dividing unit 10 (ST101), and then the representative value of each block is determined from the value of each pixel in the block. The calculation process is performed by the block representative value calculation unit 11 (ST102), and then the process of reducing the number of gradations of the representative value of each block is performed by the gradation number reduction unit 12 (ST103). The representative value of each block whose number has been reduced is stored in the block representative value storage unit 13.

  Next, the local histogram generator 14 performs processing for generating the local histogram data indicating the frequency for each gradation by counting the representative values of the blocks included in the filter region centered on the target block for each gradation. Then, the process of shaping the histogram data is performed by the histogram shaping unit 15 (ST105).

  Next, a process for generating a gradation conversion curve for each block based on the shaped histogram data is performed by the gradation conversion curve generation unit 16 (ST106), and then the gradations of a plurality of blocks around the pixel of interest are processed. A process for obtaining a plurality of gradation conversion results for the pixel of interest based on the conversion curve is performed in the gradation conversion unit 17 (ST107), and then the plurality of gradation conversion results are linearly interpolated to perform gradation of each pixel. Is obtained by the linear interpolation unit 18 (ST108).

  Next, for each pixel, processing for calculating the ratio between the gradation after gradation conversion acquired by the linear interpolation unit 18 and the gradation at the time of input from the preprocessing unit 5 is performed by the ratio calculation unit 19 ( (ST109) Then, gradation correction processing for multiplying the luminance signal and the color difference signal generated by the YC separation unit 6 by the ratio obtained by the ratio calculation unit 19 is performed by the multipliers 21 and 22 (ST110).

  Next, processing performed in each unit of the gradation correction unit 7 will be described in detail.

  FIG. 4 is an explanatory diagram for explaining a block division process performed by the block division unit 10 shown in FIG. The block dividing unit 10 performs processing for dividing the input image into a plurality of rectangular blocks. In the example shown in FIG. 4, an image having 640 pixels in width and 480 pixels in height is divided into square blocks of 8 pixels in both length and width, and the total number of blocks is 80 × 60 = 4800.

  In the present embodiment, the illustrated X direction will be described as a horizontal direction and the Y direction as a vertical direction. In addition, regarding the position of the block, the X direction is the arrangement direction of the column i, the Y direction is the arrangement direction of the row j, and the block located in the i column and the jth row is described as (i, j).

  In the block representative value calculation unit 11 illustrated in FIG. 1, a process of calculating the representative value of each block from the value of each pixel in each block set by the block dividing unit 10 is performed. The representative value calculated here is, for example, the average value or the median value of the luminance of each pixel in the block.

  The gradation number reduction unit 12 shown in FIG. 1 performs processing (dithering) to reduce the number of gradations of the representative value of each block calculated by the block representative value calculation unit 11 using an error diffusion method or the like. . This gradation number reduction process is to reduce the number of gradations suitable for generating histogram data, and the representative value of the block is converted from, for example, 256 gradations (8 bits) to 16 gradations (4 bits). .

  The block representative value storage unit 13 illustrated in FIG. 1 stores the representative value of each block having the reduced number of gradations acquired by the gradation number reducing unit 12. The block representative value storage unit 13 stores a representative value of a block for one frame.

  FIG. 5 is an explanatory diagram for explaining a local histogram generation process performed by the local histogram generator 14 shown in FIG. The local histogram generation unit 14 performs processing to acquire the representative value of each block from the block representative value storage unit 13 and generate local histogram data for each block.

  In this local histogram generation process, as shown in FIG. 5A, a filter area of a predetermined size (for example, 15 × 15) is set around the block of interest, and the representative value of the block in this filter area Are collected for each gradation, and as shown in FIG. 5B, local histogram data indicating the frequency (number of blocks) for each gradation is generated.

  In the histogram shaping unit 15 shown in FIG. 1, a histogram such as a distribution process that distributes the frequency of the required gradation to the upper and lower gradations for the local histogram data of each block acquired by the local histogram generation unit 14. A shaping process is performed.

  FIG. 6 is an explanatory diagram for explaining the gradation conversion curve generation processing performed by the gradation conversion curve generation unit 16 shown in FIG. The gradation conversion curve generation unit 16 performs a process of generating a gradation conversion curve (tone curve) of each block based on the histogram data of each shaped block acquired by the histogram shaping unit 15.

  In this gradation conversion curve generation processing, the frequency (number of blocks) for each gradation in the histogram data shown in FIG. 5B is accumulated in order from the lower gradation as shown in FIG. 6A. Generate a cumulative histogram. Then, by connecting rectangular diagonal lines defined by the frequency of each gradation on the cumulative histogram, a polygonal gradation conversion curve is obtained as shown in FIG. 6B. At this time, normalization processing is performed so that the frequency of the highest gradation in the cumulative histogram becomes the number of gradations.

  FIG. 7 is an explanatory diagram for explaining the gradation conversion process performed by the gradation conversion unit 17 and the linear interpolation process performed by the linear interpolation unit 18 shown in FIG. The gradation conversion unit 17 converts the gradation of each pixel based on the gradation conversion curves of a plurality of blocks in the vicinity of the target pixel among the gradation conversion curves of each block acquired by the gradation conversion curve generation unit 16. Processing is performed.

  In particular, in the present embodiment, four blocks, that is, a block (i, j) including the target pixel, a block (i-1, j) adjacent to the block (i-1, j), a block (i, j-1), and Based on the gradation conversion curve of the block (i-1, j-1), gradation conversion of the target pixel is performed. As a result, four gradation conversion results relating to the pixel of interest, that is, gradation L (i, j), gradation L (i-1, j), gradation L (i, j-1), and gradation L ( i-1, j-1) is obtained.

  In the linear interpolation unit 18 illustrated in FIG. 1, a process for obtaining one gradation for each pixel by linearly interpolating a plurality of gradation conversion results obtained by the gradation conversion unit 17 is performed. This linear interpolation processing is performed based on the distance from the target pixel to the center of gravity of each block. That is, the tone of the pixel of interest is obtained by a weighted average (weighted average) based on distance.

  Note that the gradation conversion curve of each block near the target pixel is weighted according to the distance from the target pixel to the center of gravity of each block to generate a gradation conversion curve related to the target pixel. The tone conversion of the pixel of interest may be performed based on the tone conversion curve.

  Next, the local histogram generation process performed by the local histogram generation unit 14 shown in FIG. 1 will be described in detail. FIG. 8 is a block diagram illustrating a schematic configuration of the local histogram generation unit 14.

  The local histogram generation unit 14 includes a weightless histogram generation unit 31, a weight coefficient setting unit 32, and a weighted histogram generation unit 33.

  The unweighted histogram generation unit 31 performs processing for acquiring the block representative value with the reduced number of gradations from the block representative value storage unit 13 and generating unweighted histogram data for each block. The weighting factor setting unit 32 performs processing for setting a weighting factor. The weighted histogram generation unit 33 performs processing for generating weighted histogram data based on the weightless histogram data acquired by the weightless histogram generation unit 31 and the weighting factor set by the weighting factor setting unit 32.

  Next, a weighted histogram generation process performed by the weighted histogram generation unit 33 will be described. FIG. 9 is an explanatory diagram for explaining the weighting region and the weighting coefficient. FIG. 10 is an explanatory diagram for explaining a procedure for generating a weighted histogram.

  In the weighted histogram generation process, as shown in FIG. 9, the filter area is divided into a plurality of weighted areas centering on the block of interest, and different weighting factors are set for each weighted area. In the present embodiment, the first to third weighting regions are set around the block of interest. The sizes of the first to third weighted areas are, for example, D1 = 15, D2 = 9, and D3 = 3. Weighting factors W1 to W3 are set in the first to third weighting regions, respectively, and the weighting factors W1 to W3 increase in order from the outermost first weighting region to the inner second and third regions. For example, W1 = 3, W2 = 5, and W3 = 6 are set.

  Here, conventionally, when the representative values of the blocks included in the filter area are aggregated for each gradation, the weighting factor set in the weighting area to which the block belongs is multiplied by the frequency (number of blocks) and aggregated. It was.

  In contrast, in the present embodiment, as shown in FIG. 10, a first filter region in which all weighting regions are integrated, and second and second filter regions in which the outer weighting regions are excluded in order from the first filter region. 3 filter regions are set, and unweighted histogram data H1, H2, and H3 are generated for the first to third filter regions, respectively.

  Specifically, as shown in FIG. 10A, unweighted histogram data H1 (i, j) for the first filter region obtained by integrating the first to third weighting regions is generated. Further, as shown in FIG. 10B, unweighted histogram data H2 (i, j) is generated for the second filter region obtained by integrating the second and third weighting regions. Further, as shown in FIG. 10C, unweighted histogram data H3 (i, j) for the third filter region based only on the third weighting region is generated.

  Then, the obtained first to third unweighted histogram data H1 (i, j), H2 (i, j) and H3 (i, j) are integrated to generate weighted histogram data. In the first to third filter regions, since the weighting regions overlap each other, instead of the first to third weighting factors W1 to W3, the first to third correction weights considering the overlapping of the weighting regions. The coefficients W1 ′, W2 ′, and W3 ′ are respectively multiplied by the first to third unweighted histogram data H1 (i, j), H2 (i, j), and H3 (i, j).

  That is, the first correction weighting coefficient W1 ′ multiplied by the first histogram data H1 (i, j) remains the first weighting coefficient W1 (= 3), but the second histogram data H2 (i, j, The second modified weighting factor W2 ′ multiplied by j) is a value (= 2) obtained by subtracting the overlap 3 from the second weighting factor W2 (= 5), and the third histogram data H3 (i, j) The third modified weighting factor W3 ′ multiplied by is a value (= 1) obtained by subtracting the overlapping 5 from the third weighting factor W3 (= 6).

  In the present embodiment, integrating two histogram data means adding the frequency in each histogram data for each gradation. Also, multiplying histogram data by a coefficient means multiplying the frequency for each gradation in the histogram data by a coefficient.

  The overall procedure of the local histogram generation process as described above will be described. FIG. 11 is a flowchart showing a procedure of local histogram generation processing performed by the local histogram generation unit 14.

  In the local histogram generation processing, first, the weightless histogram generation processing is repeated a plurality of times (here, three times) by the number of weighting factors by changing the size of the filter region in the weightless histogram generation unit 31 (ST201 to ST203). . Then, processing for setting a weighting factor is performed in the weighting factor setting unit 32 (ST204), and processing for generating weighted histogram data based on the weighting factor is performed in the weighted histogram generating unit 33 (ST205).

  Next, the unweighted histogram generation process performed by the unweighted histogram generation unit 31 shown in FIG. 8 will be described in detail. FIG. 12 is a block diagram illustrating a schematic configuration of the unweighted histogram generation unit 31. FIG. 13 is an explanatory diagram for explaining an overview of the unweighted histogram generation process. FIG. 14 is a flowchart showing the procedure of the unweighted histogram generation process.

  In the following description, the unweighted histogram generation process related to one filter area will be described, but this process is repeatedly performed while changing the size of the filter area as described above.

  As shown in FIG. 12, the unweighted histogram generation unit 31 includes a first region histogram generation unit 41, a second region histogram generation unit 42, a third region histogram generation unit 43, and a fourth region histogram generation unit 44. The horizontal direction primary difference storage unit 45 and the histogram storage unit 46 are provided.

  As shown in FIG. 13, the first region histogram generation unit 41 performs a first region histogram generation process for only the start block (0, 0) that is the start position located in the upper left corner of the image.

  In the second region histogram generation unit 42, blocks in the region excluding the start block in one line (start row) extending in the horizontal direction (first direction) from the start block, that is, blocks (1, 0) to (I, A second region histogram generation process targeting 0) is performed. In the second region histogram generation process, a process of acquiring the histogram of each block in order while shifting one by one in the horizontal direction is performed.

  In the third region histogram generation unit 43, blocks in the region excluding the start block in one line (start column) extending in the vertical direction (second direction) from the start block, that is, blocks (0, 1) to (0, A third region histogram generation process for J) is performed. In the third region histogram generation process, a process of acquiring the histogram of each block in order while shifting one by one in the vertical direction is performed.

  The fourth region histogram generation unit 44 performs a fourth region histogram generation process for blocks other than the start row and start column, that is, (1, 1) to (I, J). In the fourth region histogram generation processing, processing for one line for sequentially obtaining the histogram of each block while shifting one by one in the horizontal direction is repeatedly performed on each line (row) in order from the top.

  The histogram generation processing performed in each of the first region histogram generation unit 41, the second region histogram generation unit 42, the third region histogram generation unit 43, and the fourth region histogram generation unit 44 is sequentially performed as shown in FIG. Performed (ST301 to ST304). However, the second region histogram generation process and the third region histogram generation process may be performed in the reverse order.

  In the horizontal direction primary difference storage unit 45 shown in FIG. 12, a memory area Md for storing the horizontal direction primary difference value Dx is secured for each block. As will be described later, the horizontal direction primary difference value Dx is used only for processing for the next line in the vertical direction, and therefore the horizontal direction primary difference storage unit 45 has a memory area Md for one line. Should be secured. The histogram storage unit 46 reserves a memory area Mh for storing the histogram data H for each block.

  Next, a first region histogram generation process performed by the first region histogram generation unit 41 illustrated in FIG. 12 will be described. FIG. 15 is an explanatory diagram for explaining an overview of the first region histogram generation process. FIG. 16 is a flowchart showing the procedure of the first region histogram generation process.

  In the first region histogram generation process, histogram data of the start block (0, 0) at the upper left corner is acquired. In this first region histogram generation processing, representative values of all blocks included in the filter region of the start block (0, 0) are acquired from the block representative value storage unit 13, and the representative values of the blocks are obtained for each gradation. The histogram data H (0, 0) of the start block is obtained by aggregation (ST401 in FIG. 16). The histogram data H (0, 0) of the start block is stored in the corresponding memory area Mh (0, 0) of the histogram storage unit 46 (ST402 in FIG. 16).

  Next, the second region histogram generation process performed by the second region histogram generation unit 42 shown in FIG. 12 will be described. FIG. 17 is an explanatory diagram for explaining the outline of the second region histogram generation processing.

  As shown in FIG. 13, in the second region histogram generation process, a block-by-block histogram is acquired for blocks in the start row excluding the start block, that is, blocks (1, 0) to (I, 0). In the second region histogram generation processing, processing for acquiring a histogram for each block in order while shifting one by one in the horizontal direction is performed.

  Here, as shown in FIG. 17, when comparing the filter area of the target block (i, 0) with the filter area of the previous block (i-1, 0) in the horizontal direction, most of each filter area is Although overlapped, the leftmost column in the filter area of the previous block (i-1, 0) in the horizontal direction is out of the filter area of the target block (i, 0), and the target block (i, 0) The rightmost column is newly added to the filter area.

  Therefore, in the filter area of the previous block (i-1, 0) in the horizontal direction, the leftmost one column that is out of the filter area of the target block (i, 0) is set as the horizontal primary difference area for the decrease. In addition, the rightmost one column newly added to the filter area of the block of interest (i, 0) is included in the reduced horizontal direction primary difference area as the increased horizontal direction primary difference area. The horizontal direction primary difference data Dx (i, 0) is obtained by summing up the representative values of the blocks to be classified for each gradation, and the horizontal direction primary difference data Dx (i, 0) By integrating with the histogram data H (i-1, 0) of the block, the histogram data H (i, 0) of the block of interest can be obtained.

  In the present embodiment, the primary difference data indicates the frequency for each gradation as in the case of the histogram data, and the integration of the primary difference data and the histogram data means that the frequency in each data is a scale. Addition / subtraction is performed for each key, and addition is performed for the increased primary difference data, and subtraction is performed for the decreased primary difference data.

  Here, the horizontal primary difference area differs depending on the position of the block of interest. That is, as shown in FIG. 17A, when the block of interest is on the left side, only an increase exists and no decrease exists. Further, as shown in FIG. 17B, both the decrease and the increase exist at the intermediate position in the horizontal direction of the target block. Further, as shown in FIG. 17C, when the target block is located on the right side, only the decrease amount exists and the increase amount does not exist.

  Next, the configuration of the second region histogram generation unit 42 will be described. As shown in FIG. 12, the second region histogram generation unit 42 includes a horizontal primary difference acquisition unit 51 and an integration unit 52.

  In the horizontal direction primary difference acquisition unit 51, as described above, the filter area of the block of interest (i, 0) and the filter area of the previous block (i-1, 0) in the horizontal direction do not overlap. The representative value of the block included in the horizontal direction primary difference area for the minute and the increase is acquired from the block representative value storage unit 13 and the horizontal direction primary difference data Dx (i, 0) is acquired.

  In the integration unit 52, the horizontal direction primary difference data Dx (i, 0) acquired by the horizontal direction primary difference acquisition unit 51 and the histogram data H () of the block immediately preceding in the horizontal direction acquired from the histogram storage unit 46 ( i-1, 0) is integrated to generate histogram data H (i, 0) of the block of interest.

  Next, the procedure of the second region histogram generation process will be described. FIG. 18 is a flowchart showing the procedure of the second region histogram generation process.

  In the second region histogram generation process, first, the column i is initialized (ST501). Thereby, the histogram generation processing for each block is started from the block (1, 0).

  In the histogram generation process for each block, first, the horizontal primary difference acquisition unit 51 acquires horizontal primary difference data Dx (i, 0) (ST502), and the horizontal primary difference data Dx ( i, 0) is stored in the corresponding memory area Md (i, 0) of the horizontal primary difference storage unit 45 (ST503). Next, the horizontal primary difference data Dx (i, 0) is integrated into the histogram data H (i-1,0) of the previous block in the horizontal direction, and the histogram data H (i, 0) of the block of interest. ) Is acquired (ST504), and the histogram data H (i, 0) of the block of interest is stored in the corresponding memory area Mh (i, 0) of the histogram storage unit 46 (ST505).

  In this way, when the histogram generation processing of one block is completed, the column i is incremented by 1 (ST506), and the column i exceeds the maximum value I, that is, all processing of the target block is completed. Is determined (ST507), and if all of the processing of the target block is not completed (No in ST507), the process proceeds to the processing of the next block until all of the processing of the target block is completed. Repeated.

  As described above, in the second region histogram generation process, the representative values of the blocks included in the horizontal primary difference region in which the filter region of the target block and the filter region of the previous block in the horizontal direction do not overlap are totaled. The obtained horizontal direction primary difference data is obtained, and the horizontal direction primary difference data and the histogram data of the previous block in the horizontal direction are integrated to generate the histogram data of the block of interest. As a result, it is not necessary to aggregate the representative values of all the blocks included in the filter area, so that the calculation burden can be reduced.

  Next, a third region histogram generation process performed by the third region histogram generation unit 43 illustrated in FIG. 12 will be described. FIG. 19 is an explanatory diagram for explaining the outline of the third region histogram generation processing.

  As shown in FIG. 13, in the third region histogram generation process, histogram data of each block is acquired for the blocks in the start column excluding the start block, that is, the blocks (0, 1) to (0, J). . In the third region histogram generation processing, processing for acquiring histogram data of each block in order in the vertical direction is performed.

  Here, as shown in FIG. 19, when comparing the filter area of the target block (0, j) with the filter area of the previous block (0, j-1) in the vertical direction, most of the filter areas are Although overlapping, the uppermost row in the filter area of the immediately preceding block (0, j-1) in the vertical direction is out of the filter area of the target block (0, j), and the target block (0, j) The lowermost row is newly added to the filter area.

  Therefore, in the filter area of the previous block (0, j-1) in the vertical direction, the uppermost one line that is out of the filter area of the target block (0, j) is defined as the vertical difference area in the vertical direction. In addition, the lowermost row newly added to the filter area of the target block (0, j) is set as the vertical direction primary difference area for the increase, and the vertical direction primary difference area for the decrease and the increase The representative values of the included blocks are aggregated for each gradation to obtain the vertical direction primary difference data Dy (0, j), and this vertical direction primary difference data Dy (0, j) is one before in the vertical direction. The histogram data H (0, j) of the block of interest can be obtained by integrating with the histogram data H (0, j-1) of the block.

  Here, the vertical direction primary difference area differs depending on the position of the target block. That is, as shown in FIG. 19 (A), at the position where the target block is on the upper side, there is only an increase and no decrease. Further, as shown in FIG. 19B, both the decrease amount and the increase amount exist at the position where the target block is in the middle in the vertical direction. Further, as shown in FIG. 19C, at the position where the block of interest is at the lower side, there is only a decrease and no increase.

  Next, the configuration of the third region histogram generation unit 43 will be described. As illustrated in FIG. 12, the third region histogram generation unit 43 includes a vertical direction primary difference acquisition unit 53 and an integration unit 54.

  In the vertical direction primary difference acquisition unit 53, as described above, the filter region of the target block (0, j) and the filter region of the previous block (0, j-1) in the vertical direction do not overlap. The representative values of the blocks included in the vertical direction primary difference area for the minute and the increase are acquired from the block representative value storage unit 13 to acquire the vertical direction primary difference data Dy (0, j).

  In the integration unit 54, the vertical direction primary difference data Dy (0, j) acquired by the vertical direction primary difference acquisition unit 53 and the histogram data H () of the previous block in the vertical direction acquired from the histogram storage unit 46 ( 0, j-1) is integrated to generate histogram data H (0, j) of the block of interest.

  Next, the procedure of the third region histogram generation process will be described. FIG. 20 is a flowchart showing the procedure of the third region histogram generation process.

  In the third region histogram generation process, first, row j is initialized (ST601). Thereby, the histogram generation processing for each block is started from the block (0, 1).

  In the histogram generation process for each block, first, the vertical primary difference acquisition unit 53 acquires vertical primary difference data Dy (0, j) (ST602). Next, the vertical difference data Dy (0, j) in the vertical direction is integrated with the histogram data H (0, j-1) of the previous block in the vertical direction, so that the histogram data H (0, j of the block of interest). ) Is acquired (ST603), and the histogram data H (0, j) of the target block is stored in the corresponding memory area Mh (0, j) of the histogram storage unit 46 (ST604).

  When the histogram generation processing for one block is completed in this way, the row j is incremented by 1 (ST605), and then the row j exceeds the maximum value J, that is, all the processing of the target block is completed. Is determined (ST606), and if all the processing of the target block is not completed (No in ST606), the process proceeds to the next block, and is repeated until all the target blocks are completed. .

  As described above, in the third region histogram generation process, the representative values of the blocks included in the vertical direction primary difference region in which the filter region of the target block and the filter region of the previous block in the vertical direction do not overlap are totaled. The vertical direction primary difference data is obtained, and the vertical direction primary difference data and the histogram data of the previous block in the vertical direction are integrated to generate the histogram data of the block of interest. As a result, it is not necessary to aggregate the representative values of all the blocks included in the filter area, so that the calculation burden can be reduced.

  Next, a fourth region histogram generation process performed by the fourth region histogram generation unit 44 illustrated in FIG. 12 will be described. 21, FIG. 22, and FIG. 23 are explanatory diagrams for explaining the outline of the fourth region histogram generation processing. FIG. 24 is an explanatory diagram for explaining the fourth region histogram generation processing in detail.

  As shown in FIG. 13, in the fourth region histogram generation process, histogram data of each block is acquired for blocks other than the start row and start column, that is, (1, 1) to (I, J). In the fourth region histogram generation processing, processing for one line for acquiring histogram data of each block in order in the horizontal direction is repeatedly performed on each line (row) in order from the top.

  In this fourth area histogram generation process, as shown in FIGS. 21, 22, and 23, there are nine patterns according to the position of the block of interest. First, the representative pattern shown in FIG. This will be described with reference to FIG.

  First, as shown in FIG. 24 (A), the filter region of the target block (i, j) and the filter region of the immediately preceding block (i-1, j) do not overlap with each other in the horizontal direction. If the horizontal direction primary difference data Dx (i, j) of the difference area is obtained, the horizontal direction primary difference data Dx (i, j) is converted into the histogram data H (i−1) of the immediately previous block in the horizontal direction. , J), the histogram data H (i, j) of the block of interest can be obtained. This is the same as the second region histogram generation process.

  On the other hand, in the fourth area histogram generation process, the horizontal direction primary difference data (i, 0) of the block of the start row acquired in the second area histogram generation process or the vertical direction acquired in the fourth area histogram generation process. By using the horizontal primary difference data (i, j-1) of the previous block, the horizontal primary differential data Dx (i, j) of the block of interest can be obtained with a small amount of calculation.

  That is, as shown in FIGS. 24B-1 and 24B-2, the horizontal primary difference area of the block of interest (i, j) and the previous block (i, j-1) in the vertical direction. When compared with the horizontal direction primary difference area, most of the horizontal direction primary difference areas overlap, but the uppermost in the horizontal direction primary difference area of the previous block (i, j-1) in the vertical direction. Are out of the horizontal primary difference area of the target block (i, j), and the two lowest blocks in the horizontal primary difference area of the target block (i, j) are newly added. .

  Therefore, the uppermost two blocks that deviate from the horizontal primary difference area of the block of interest (i, j) in the horizontal primary differential area of the previous block (i, j-1) in the vertical direction are reduced. Minute horizontal direction secondary difference block (secondary difference region), and the lowest two blocks newly added to the horizontal direction primary difference region of the target block (i, j) As the secondary difference block (secondary difference area), horizontal secondary difference data D2x (i, j) based on the representative value of the horizontal secondary difference block for the decrease and increase is obtained, and the horizontal secondary difference is obtained. By integrating the data D2x (i, j) with the horizontal primary difference data Dx (i, j-1) of the previous block in the vertical direction, the horizontal primary difference data Dx (i of the block of interest , J).

  When the horizontal direction primary difference data Dx (i, j) of the block of interest is thus obtained, the horizontal direction primary difference data Dx (i, j) is converted to the block (i− 1, j) histogram data H (i, j) of the block of interest (i, j) can be obtained by integrating with the histogram data H (i-1, j).

  In the present embodiment, the secondary difference data indicates the frequency for each gradation as in the histogram data, and the integration of the secondary difference data and the primary difference data means the frequency in each data. Is added / subtracted for each gradation, and addition is performed for the secondary difference data for the increase, and subtraction is performed for the secondary difference data for the decrease.

  Here, as shown in FIG. 24C, the horizontal second-order differential block exists outside or inside the four corners of the filter area of the block of interest (i, j), but is located in the lower right direction. Since the secondary difference block is newly added to the incremented lateral primary difference area, it becomes an incremented lateral secondary difference block, and the lateral secondary difference block located at the upper right is the incremented lateral direction 1. Since it is out of the next difference area, it becomes a reduced horizontal secondary difference block, and the horizontal second difference block located at the lower left is newly added to the reduced horizontal primary difference area, so the reduced horizontal direction The secondary difference block, which is the secondary difference block and located in the upper left, deviates from the reduced horizontal primary difference area, and thus becomes the increased secondary differential block.

  Further, the horizontal secondary difference block differs depending on the position of the target block, as shown in FIGS. That is, as shown in FIG. 21 (A), at the position where the target block is on the upper side and the left side, there is only one on the lower right side, and as shown in FIG. 21 (B), the target block is on the upper side and in the horizontal direction. In the middle position, there are two on the left and right sides of the lower side, and as shown in FIG. 21C, there is only one on the lower left side when the target block is on the upper side and on the right side.

  As shown in FIG. 22 (A), when the block of interest is in the middle and leftward position in the vertical direction, there are two blocks on the right and up, and as shown in FIG. 22 (B), the block of interest is in the middle of the vertical direction. At the middle position in the horizontal direction, there are four on the left and right and up and down, and as shown in FIG. 22C, there are two blocks of interest on the left and top at the middle and right side in the vertical direction.

  As shown in FIG. 23 (A), when the target block is at the lower and left positions, there is only one at the upper right. As shown in FIG. 23 (B), the target block is at the lower side and in the middle in the horizontal direction. In the position, there are two on the left and right on the upper side, and as shown in FIG. 23C, there is only one on the upper left in the position where the target block is at the lower side and the right side.

  Next, the configuration of the fourth region histogram generation unit 44 will be described. As shown in FIG. 12, the fourth region histogram generation unit 44 includes a horizontal direction secondary difference acquisition unit 55, a horizontal direction primary difference acquisition unit 56, and an integration unit 57.

  The horizontal direction secondary difference acquisition unit 55 acquires the representative value of the horizontal direction secondary difference block from the block representative value storage unit 13, and generates the horizontal direction secondary difference data D2x (i, j) of the block of interest.

  The horizontal direction primary difference acquisition unit 56 acquires the horizontal direction primary difference data Dx (i, j-1) of the immediately preceding block in the vertical direction from the horizontal direction primary difference storage unit 45, and this horizontal direction The primary difference data Dx (i, j-1) and the horizontal secondary difference data D2x (i, j) acquired by the horizontal secondary difference acquisition unit 55 are integrated, and the horizontal primary of the target block is integrated. Difference data Dx (i, j) is generated.

  In the integration unit 57, the horizontal primary difference data Dx (i, j) of the block of interest acquired by the horizontal primary difference acquisition unit 56 and the histogram of the block immediately preceding in the horizontal direction acquired from the histogram storage unit 46 are obtained. The data H (i-1, j) is integrated to generate histogram data H (i, j) of the block of interest.

  Next, the procedure of the fourth region histogram generation process will be described. FIG. 25 is a flowchart showing the procedure of the fourth region histogram generation process.

  In the fourth region histogram generation process, first, column i and row j are initialized (ST701). Thereby, the histogram generation processing for each block is started from the block (1, 1).

  In the histogram generation processing for each block, first, the horizontal secondary difference acquisition unit 55 acquires the horizontal secondary difference data D2x (i, j) (ST702). Next, the horizontal primary difference data D2x (i, j) is integrated into the horizontal primary difference data Dx (i, j-1) of the previous block by the horizontal primary difference acquisition unit 56. The horizontal primary difference data Dx (i, j) is acquired (ST703), and the horizontal primary difference data Dx (i, j) is stored in the corresponding memory area Md (i , J) (ST704).

  Then, the horizontal primary difference Dx (i, j) is integrated with the histogram data H (i-1, j) of the previous block in the horizontal direction, and the histogram data H (i, j) of the block of interest is obtained. Obtained (ST705), the histogram data H (i, j) of this block of interest is stored in the corresponding memory area Mh (i, j) of the histogram storage unit 46 (ST706).

  When the histogram generation processing for one block is completed in this way, the column i is incremented by 1 (ST707), and the column i exceeds the maximum value I, that is, whether or not the processing for one row is completed. (ST708) If the process for one row is not completed (No in ST708), the process proceeds to the next block, and is repeated until the process for the block for one line is completed. When the block processing for one row is completed (Yes in ST708), the row j is incremented by 1 (ST709), and the row j exceeds the maximum value J, that is, the processing for all the rows is completed. Is determined (ST 710), and if all the lines have not been processed (NO in ST 710), the process proceeds to the next line, and is repeated until all the lines have been processed.

  As described above, in the fourth region histogram generation process, the horizontal direction primary difference region of the block of interest and the horizontal direction primary difference region of the previous block in the vertical direction are not overlapped. By obtaining horizontal secondary difference data from the representative value and integrating the horizontal secondary difference data and the horizontal primary difference data of the previous block in the vertical direction, the horizontal primary difference of the block of interest is integrated. Data is acquired, and then the acquired horizontal primary difference data of the block of interest and the histogram data of the previous block in the horizontal direction are integrated to generate histogram data of the block of interest. As a result, it is not necessary to aggregate the representative values of all the blocks included in the filter area, and furthermore, the horizontal primary difference data of the block of interest can be obtained with a small amount of calculation, thereby reducing the calculation burden. it can.

  As shown in FIG. 17, in the second area histogram generation process, there are three appearance patterns of the horizontal direction primary difference area. As shown in FIG. 19, in the third area histogram generation process, the vertical direction There are three appearance patterns of the primary difference area. As shown in FIGS. 21 to 23, in the fourth area histogram generation process, nine appearance patterns of the horizontal direction primary difference area and the horizontal direction secondary difference block are generated. However, these appearance patterns differ depending on the size of the filter area for the entire image.

  For example, if the number of blocks in the vertical direction of the filter area is larger than the number of vertical blocks in the entire image, as shown in FIG. In the fourth region histogram generation process, as shown in FIG. 22, there is no pattern in which both of the first-order difference areas in the vertical direction are present at the same time. A pattern in which both directional secondary difference blocks exist simultaneously does not occur.

  As mentioned above, although this invention was demonstrated based on specific embodiment, these embodiment is an illustration to the last, This invention is not limited by these embodiment, The range which does not deviate from the meaning of this invention It can be changed as appropriate.

  That is, in this embodiment, tone correction is performed based on histogram data for the purpose of enhancing contrast. However, the present invention is not limited to this, and image correction is performed based on histogram data. It can be widely applied to image processing.

  In the present embodiment, the image is divided into a plurality of blocks and the histogram data for each block is generated. However, the present invention is not limited to this, and a configuration for generating histogram data for each pixel. It can also be applied to.

  In the present embodiment, the fourth region histogram generation unit 44 repeats processing for one line for acquiring histogram data of each block in order in the horizontal direction (first direction) for each line (row) in order from the top. The secondary difference acquisition unit acquires the secondary difference data in the vertical direction (second direction). However, the vertical and horizontal directions are reversed, that is, the first direction is the vertical direction, A configuration in which the direction 2 is the horizontal direction is also possible.

  The image processing apparatus and the image processing method according to the present invention have an effect of speeding up the process of generating histogram data indicating the frequency for each gradation, and an image for correcting an image based on the histogram data It is useful as a processing device and an image processing method.

7 Tone correction unit (image processing device)
10 block division unit 11 block representative value calculation unit 13 block representative value storage unit 14 local histogram generation unit 16 gradation conversion curve generation unit 17 gradation conversion unit 31 weightless histogram generation unit 32 weight coefficient setting unit 33 weighted histogram generation Unit 41 first region histogram generation unit 42 second region histogram generation unit 43 third region histogram generation unit 44 fourth region histogram generation unit 45 horizontal direction primary difference storage unit 46 histogram storage unit 51 horizontal direction primary difference acquisition unit 52 Integration unit 53 Vertical primary difference acquisition unit 54 Integration unit 55 Horizontal secondary difference acquisition unit 56 Horizontal primary difference acquisition unit 57 Integration unit

Claims (7)

An image processing apparatus that generates histogram data for a filter region set around a target position on an input image and corrects the image based on the histogram data,
A local histogram generator that sequentially generates the histogram data at each position while shifting the target position one by one on the image;
This local histogram generator is
A histogram storage for storing the histogram data at each position on the image;
A primary difference acquisition unit that acquires primary difference data for a primary difference area in which the filter area at a position of interest and the filter area at a position immediately before the target area do not overlap;
An integration unit that integrates the primary difference data acquired by the primary difference acquisition unit and the histogram data at the previous position stored in the histogram storage unit, and acquires histogram data at the target position. And an image processing apparatus. The local histogram generation unit performs processing for one line for sequentially generating the histogram data at each position while shifting the position of interest in the first direction, with one line in a second direction orthogonal to the first direction. It repeats while shifting one by one,
A primary difference storage unit for storing the primary difference data acquired by the primary difference acquisition unit;
Secondary difference data for acquiring secondary difference data for a secondary difference area in which the primary difference data at the position of interest and the primary difference data at the position immediately before in the second direction do not overlap. An acquisition unit,
The primary difference acquisition unit includes the secondary difference data acquired by the secondary difference acquisition unit and the primary difference at a position immediately before in the second direction stored in the primary difference storage unit. The image processing apparatus according to claim 1, wherein the primary difference data at the position of interest is acquired by integrating data. The local histogram generator is
A first region histogram generator for a start position located at a corner of the image;
A second region histogram generator for a region excluding the start position in one line extending in the first direction from the start position;
A third region histogram generator for a region excluding the start position in one line extending in the second direction from the start position;
A fourth region histogram generation unit for a region excluding a target region of the first region histogram generation unit, the second region histogram generation unit, and the third region histogram generation unit;
The first region histogram generation unit obtains the histogram data by summing feature values at each position in the filter region at the start position,
The second region histogram generation unit includes the primary difference acquisition unit and the integration unit, and shifts the target position in the first direction from a position shifted by one in the first direction with respect to the start position. Generate the histogram data sequentially while shifting,
The third region histogram generation unit includes the primary difference acquisition unit and the integration unit, and moves a target position in a second direction from a position shifted by one in a second direction with respect to the start position. Generate the histogram data sequentially while shifting,
The fourth region histogram generation unit includes the secondary difference acquisition unit, the primary difference acquisition unit, and the integration unit, one each in a first direction and a second direction with respect to the start position. The processing for one line for sequentially generating the histogram data at each position while shifting the target position in the first direction from the shifted position is repeated while shifting one line in the second direction. Item 3. The image processing apparatus according to Item 2. Gradation correction is performed on the image,
The image correction unit
A gradation conversion curve generation unit that generates a gradation conversion curve based on the histogram data;
The image processing apparatus according to claim 1, further comprising: a gradation conversion unit that performs gradation conversion on each pixel using the gradation conversion curve. A block dividing unit for dividing the image into a plurality of blocks;
A block representative value obtaining unit for obtaining a representative value for each block from the value of each pixel in the block;
The image processing apparatus according to claim 1, wherein the local histogram generation unit generates histogram data for each block. The local histogram generator generates weighted histogram data that is weighted according to the distance from the position of interest,
A weightless histogram generator for generating weightless histogram data for a plurality of the filter regions having different sizes set with different weighting factors;
A weighted histogram generating unit configured to generate weighted histogram data by integrating the weightless histogram data for each of the plurality of filter regions by multiplying the weighted coefficient data by each of the weighted histogram data. The image processing apparatus according to claim 1. An image processing method for generating histogram data for a filter region set around a target position on an input image and correcting the image based on the histogram data,
Sequentially generating the histogram data at each position while shifting the position of interest on the image one by one,
The step of generating the histogram data includes:
Obtaining primary difference data targeted for a primary difference area in which the filter area at the position of interest and the filter area at the previous position do not overlap; and
Integrating the primary difference data and the histogram data at the previous position to obtain histogram data at the position of interest, and an image processing method.