JP5052482B2 - Blur correction processing device - Google Patents

Description

  The present invention relates to a blur correction processing apparatus that enhances the sharpness of an image.

  In an apparatus such as a copying machine or a scanner, a line image sensor is used to read a document. In these apparatuses, when the original is lifted from the reading surface, the distance between the line image sensor and the original changes, so that the focus is not achieved, and the captured image data is blurred. In particular, when a contact image sensor (CIS) is used as the line image sensor, blurring is likely to occur because the depth of focus is shallow. For this reason, in order to correct blur and improve the image quality of a captured image, sharpening processing is performed on the captured image to enhance sharpness.

  As a sharpening method, an unsharp mask has been conventionally known as disclosed in Patent Document 1. An unsharp mask emphasizes the sharpness of an image by enhancing high frequency components of the image. When an image before the unsharp mask processing (original image) is represented by I and an image after the unsharp mask processing is represented by Ir, the image Ir after the unsharp mask processing is represented by Expression (1).

  Here, in Expression (1), LP (I) represents an image obtained as a result of filtering the original image I with a low-pass filter. In Expression (1), a is a value representing the degree of sharpening by unsharp mask processing. In Expression (1), the low frequency component LP (I) is subtracted from the original image I to obtain the high frequency component (I-LP (I)) of the original image I. Further, the high frequency component (I-LP (I)) of the original image is multiplied by the sharpening degree a, and the result is added to the original image I to emphasize the high frequency component of the original image, It is sharpening.

  In Patent Document 1, an example of a moving average filter is shown as a low-pass filter for unsharp mask processing. In the moving average filter, the filter coefficient L (u, v) is represented by 1 / N and is a value independent of the positional relationship (u, v) with the target pixel. N is the total number of pixels in the moving average area.

  In Patent Document 2, an example of a Gaussian filter is shown as a low-pass filter for unsharp mask processing. In the Gaussian filter, a normal distribution type filter coefficient L (u, v) as shown in Expression (2) is used.

In Equation (2), σ ² is a parameter representing the spread of the normal distribution function.

JP 2006-246080 A (page 3, line 48 to line 50, page 4, line 1 to line 21, page 16, line 31 to line 36) JP-A-10-243238 (5th page, lines 37 to 50)

  When a moving average filter is used as a low-pass filter for unsharp mask processing, the filter coefficients are all the same value, so that the amount of computation can be reduced by performing coefficient multiplication together. There is a problem that the image quality of the converted image deteriorates.

  When a Gaussian filter is used as a low-pass filter for unsharp mask processing, the filter coefficient is a normal distribution type, so that a sharp image with good image quality can be obtained. There is a problem that the circuit scale and processing time become large.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain a blur correction processing apparatus that suppresses a reduction in image quality and suppresses a circuit scale or processing time.

  In order to solve the above-mentioned problem, a first aspect of the present invention is a pixel intensity calculation unit that calculates a pixel intensity value for each pixel data from a plurality of signal components included in each pixel data constituting image data. A filter unit that calculates a pixel intensity filter value of the target pixel from the pixel intensity value calculated by the pixel intensity calculation unit, and the pixel intensity value and the filter of the target pixel calculated by the pixel intensity calculation unit A correction coefficient calculation unit that calculates a correction coefficient of the target pixel for sharpening the image data from the pixel intensity filter value of the target pixel calculated by the unit, and a target pixel calculated by the correction coefficient calculation unit A correction unit that sharpens the pixel data of the target pixel of the image data using the correction coefficient and obtains pixel data of the sharpened image data, and the filter unit sets the target pixel in advance. A plurality of areas having different sizes are provided, and for each of the areas, the pixel intensity value calculated by the pixel intensity calculating unit is used to add the pixel intensity values of the pixel data in the area. Accordingly, a region intensity calculation unit that calculates a region intensity value of each of the regions, a region selection unit that selects one or more regions from the plurality of regions, and one or more selected by the region selection unit A region intensity filter unit that calculates a pixel intensity filter value of the pixel of interest from the region intensity value calculated by the region intensity calculation unit corresponding to the region.

  According to the first aspect of the present invention, in the region intensity calculation unit, a plurality of regions having different sizes around the pixel of interest are provided in advance, and the pixel intensity of the pixel data in that region is provided for each region. The region intensity value is calculated by adding the values, the region selecting unit selects one or more regions from the plurality of regions, and the region intensity filter unit selects the region intensity corresponding to the selected one or more regions. Since the pixel intensity filter value of the target pixel is calculated from the value, the number of times the filter coefficient is multiplied is reduced, and the amount of calculation of the filter can be reduced while suppressing deterioration in image quality, thereby reducing the circuit scale or processing time. be able to. In addition, with the above configuration, it is possible to obtain filter results with different size regions and different filter coefficients while suppressing the amount of filter calculation.

Embodiment 1 FIG.
<Overall configuration>
In the blur correction processing apparatus according to the first embodiment of the present invention, for image data in a blurred state, such as image data captured by an image sensor, a pixel intensity value corresponding to luminance is obtained from pixel data constituting the image data. Calculate, apply low-pass filter processing to the calculated pixel intensity value to calculate a pixel intensity filter value, calculate a correction coefficient for image data sharpening from the calculated pixel intensity value and pixel intensity filter value, and correct the correction Sharpen image data by coefficients.

  FIG. 1 is a block diagram showing a blur correction processing apparatus 1 according to the first embodiment. As shown in FIG. 1, the blur correction processing apparatus 1 according to Embodiment 1 includes a pixel intensity calculation unit 11, a filter unit 12, a correction coefficient calculation unit 13, a correction unit 14, and input terminals 15 and 16. And an output terminal 17.

  The input terminal 15 receives blurred image data I such as image data captured by an image sensor (not shown). Further, the input terminal 16 receives an intensity coefficient bst for adjusting the degree of sharpening determined by a predetermined control unit (not shown).

  The pixel intensity calculation unit 11 calculates a pixel intensity value FP from a plurality of signal components included in the pixel data constituting the image data I input from the input terminal 15, and supplies it to the filter unit 12 and the correction coefficient calculation unit 13. Output.

  The filter unit 12 includes the pixel intensity value FP calculated by the pixel intensity calculation unit 11 and the blurring degree Bj of the target pixel input from the terminal 27 (j: blurring degree identification number, j = 1, 2,..., M From (M: integer)), the pixel intensity filter value FI of the target pixel is calculated by low-pass filter processing, and is output to the correction coefficient calculation unit 13.

  The correction coefficient calculation unit 13 includes the pixel intensity filter value FI of the target pixel calculated by the filter unit 12, the pixel intensity value FP of the target pixel calculated by the pixel intensity calculation unit 11, and the sharpness input from the input terminal 16. A correction coefficient Ga for sharpening the image data I input to the input terminal 15 is calculated from the intensity coefficient bst for adjusting the degree of conversion, and is output to the correction unit 14.

  The correction unit 14 sharply emphasizes the sharpness of the image data I from the pixel data of the target pixel of the image data I input from the input terminal 15 and the correction coefficient Ga of the target pixel calculated by the correction coefficient calculation unit 13. Pixel data of the converted image data Is is obtained and output from the output terminal 17.

<Detailed Configuration of Pixel Intensity Calculation Unit 11>
Next, the pixel intensity calculation unit 11 will be described in detail.

  The pixel intensity calculation unit 11 calculates a pixel intensity value FP from a plurality of signal components of pixel data constituting the image data I input from the input terminal 15 and outputs it to the filter unit 12 and the correction coefficient calculation unit 13. . Examples of the plurality of signal components included in the pixel data constituting the input image data I include the R component, G component, and B component of the RGB signal. The pixel intensity value FP is calculated by calculating, for example, luminance or a value corresponding to luminance (luminance equivalent value) as the pixel intensity value FP. When a luminance equivalent value is calculated as the pixel intensity value FP, the pixel intensity value FP is calculated as in Expression (3).

  Note that RI, GI, and BI in Equation (3) represent the R component, G component, and B component of the RGB signal.

<Detailed Configuration of Filter Unit 12>
Next, the filter unit 12 will be described in detail.

  The filter unit 12 includes the pixel intensity value PF of the target pixel calculated by the pixel intensity calculation unit 11 and the blur degree Bj (j: blur degree identification number, j = 1, 2) of the target pixel input from the terminal 27. ,..., M (M: integer)), the pixel intensity filter value FI is calculated by low-pass filter processing and is output to the correction coefficient calculation unit 13.

  FIG. 2 is a block diagram of the filter unit 12. The filter unit 12 includes a region intensity calculation unit 21, a region selection unit 22, a region intensity filter unit 23, input terminals 24 and 25, and an output terminal 26.

  The pixel intensity value FP calculated by the pixel intensity calculation unit 11 is input to the input terminal 24. The input terminal 25 has a blur degree Bj (j: blur degree identification number, j = 1, 2) of the target pixel determined by a predetermined control unit (not shown) and input from the terminal 27 to the correction processing apparatus 1. ,..., M (M: integer)) are input.

  The area intensity calculation unit 21 previously provides a plurality of areas Ai (i: area number, i = 1, 2,..., N (N: integer)) having different sizes with the pixel of interest P1 as the center. 4, by using the pixel intensity value FP from the input terminal 24 and adding the pixel intensity value FP of the pixels in the area Ai for each area Ai, the area intensity value FAi of each area Ai is added. Is output to the region intensity filter unit 23. Note that FP (u, v) in the equation (4) represents the pixel intensity value of the pixel at the (u, v) coordinate.

  Each area Ai is set to have a rectangular shape (rectangular or square) having a different size, and among the areas Ai, all of the larger areas include all of the smaller areas (ie, are nested). ) May be set. FIG. 3 shows a case where, for example, six square areas A1 to A6 are provided in a nested state as each area Ai. In FIG. 3, the number of areas Ai is 6, but is generally set to an integer of 1 or more.

  The area selection unit 22 selects and selects one or more areas Ai from a plurality of areas Ai having different sizes around the target pixel P1 according to the blurring degree Bj of the target pixel P1 from the input terminal 25. The region number i of the region Ai is output to the region intensity filter unit 23 as region selection information H.

  For example, as illustrated in FIGS. 4 and 5, the region selection unit 22 associates a plurality of blur levels B1 to B6 and a plurality of regions A1 to A6 in advance, and based on the association, a plurality of regions A1 to A1. From A6, one or more areas corresponding to the degree of blur Bj of the target pixel P1 from the input terminal 25 are selected, and the area number i of the selected area Ai is output as area selection information H. 4 and 5, each blur level Bj is associated with a region circumscribing the range of the blur level Bj and a region included in the range of the blur region among the plurality of regions Ai. In FIG. 4, as an example, the number of blur levels Bj is 6 and the number of areas Ai is 6, but the present invention is not limited to this.

  The region intensity filter unit 23 performs a filtering process on the pixel intensity filter value FI of the target pixel P1 from the region intensity value FAi calculated by the region intensity calculation unit 21 and the region selection information H obtained by the region selection unit 22. Is output from the output terminal 26.

  The above filter processing is performed based on, for example, Expression (5). In Expression (5), the area intensity value FAi of each area Ai selected based on the area selection information H is multiplied by the area intensity weighting coefficient ki corresponding to each area Ai, and the sum of the multiplication results (first) 1), the area intensity weighting coefficient ki corresponding to each area Ai and the number of pixels Ni of each area Ai are multiplied, and the sum of the multiplication results (second sum) is calculated. The pixel intensity filter value FI is calculated by dividing the first sum by the second sum. Note that S in Expression 5 represents a set of region numbers i included in the region selection information H obtained by the region selection unit 22.

  A plurality of sets of area intensity weighting coefficients ki corresponding to the respective areas Ai are prepared corresponding to the respective blur levels Bj (j = 1, 2,...). FIGS. 6A to 6F are examples of ki sets, and the ki sets in FIGS. 6A to 6F correspond to B1 to B6 in FIGS. 4 and 5, respectively. The region intensity filter unit 23 selects a set of ki corresponding to the degree of blur Bj of the target pixel P1 input to the input terminal 25 based on the above correspondence from the plurality of sets of ki prepared as described above. Then, using the selected ki set, the calculation of Expression (5) is performed.

  FIG. 7 is a diagram in which the filter coefficient of each pixel is arranged at the position of each pixel in the region Ai, and FIGS. 7A to 7F show the sets of ki in FIGS. 6A to 6F, respectively. It corresponds to the case of using. The filter coefficients in FIG. 7 are values close to a Gaussian filter arranged in a square shape, and a result close to a Gaussian filter is obtained.

Note that the pixel intensity filter value FI may be calculated using Equation (6) instead of Equation (5). In Expression (6), the area intensity value FAi of each area Ai selected based on the area selection information H is divided by the number of pixels Ni of each area Ai, and the division result corresponds to each area Ai. The pixel intensity filter value FI is calculated by multiplying the predetermined region intensity weighting coefficient k ₀ i and adding all the multiplication results. Note that S in Expression 6 represents a set of region numbers i included in the region selection information H obtained by the region selection unit 22.

Note that k ₀ i in equation (6) is obtained based on equation (7). In Expression (7), a multiplication value is calculated by multiplying the area intensity weighting coefficient ki corresponding to the area Ai selected based on the area selection information H by the number of pixels Ni included in the area Ai. The region intensity weighting coefficient kj corresponding to each region Aj selected based on the selection information H is multiplied by the number of pixels Nj of each region Aj, and the sum of the multiplication results is calculated. Dividing by the sum of Note that S in Expression 7 represents a set of region numbers i included in the region selection information H obtained by the region selection unit 22.

  In Expression (6), since the area intensity value FAi of each area Ai is divided by the number of pixels Ni of each area Ai, the number of divisions is increased, but an increase in the value being calculated can be suppressed. Thus, the circuit scale can be kept small.

Further, the pixel intensity filter value FI may be calculated using Expression (8) instead of Expression (5). In Expression (8), the area intensity value FAi of each area Ai selected based on the area selection information H is multiplied by a predetermined area intensity weighting coefficient k ₁ i corresponding to each area Ai, and the multiplication is performed. The sum of the results is calculated to calculate the pixel intensity filter value FI. Note that S in Expression 8 represents a set of region numbers i included in the region selection information H obtained by the region selection unit 22.

Note that k ₁ i in equation (8) is obtained based on equation (9). In Expression (9), the region intensity weighting coefficient kj corresponding to each region Aj selected based on the region selection information H is multiplied by the number of pixels Nj of each region Aj, and the sum of the multiplication results is calculated. The area intensity weighting coefficient ki corresponding to the area Ai selected based on the area selection information H is divided by the sum. Note that S in Expression (9) represents a set of region numbers i included in the region selection information H obtained by the region selection unit 22.

  In equation (8), division at the time of FI calculation is eliminated, so that the circuit scale can be kept small.

<Detailed Configuration of Correction Coefficient Calculation Unit 13>
Next, the correction coefficient calculation unit 13 will be described in detail.

  The correction coefficient calculator 13 receives the pixel intensity filter value FI of the target pixel P1 calculated by the filter unit 12, the pixel intensity value FP of the target pixel calculated by the pixel intensity calculator 11, and the input terminal 16 A correction coefficient Ga for sharpening the image data I input from the input terminal 15 is calculated from the intensity coefficient bst for adjusting the sharpening degree, and is output to the correction unit 14. The correction coefficient Ga for sharpening the image data I is calculated by, for example, Expression (10).

  In Expression (10), the calculation is performed by using the ratio of the pixel intensity filter value FI of the target pixel to the pixel intensity value FP of the target pixel P1, and is further calculated based on the correction coefficient bst. In Expression (10), the enhancement coefficient bst is set to a value larger than 0 in order to sharpen the image data I. By increasing the enhancement coefficient bst, the degree of sharpening of the image data I can be increased.

<Detailed Configuration of Correction Unit 14>
Next, the correction unit 14 will be described in detail.

  The correction unit 14 emphasizes the sharpness of the image data I from the pixel data of the target pixel P1 of the image data I input from the input terminal 15 and the correction coefficient Ga of the target pixel P1 calculated by the correction coefficient calculation unit 13. Pixel data of the sharpened image data Is is obtained and output. The pixel data of the sharpened image data Is is calculated, for example, by multiplying the pixel data of the target pixel P1 of the image data I by the correction coefficient Ga of the target pixel P1 as in Expression (11).

  In Expression (11), RO, GO, and BO represent the R component, G component, and B component of the pixel data for the pixel of interest P1 of the sharpened image data Is, and RI, GI, and BI represent the attention of the image data I. Represents the R component, G component, and B component of the pixel P1.

  In the above description, the pixel intensity calculation unit 11 calculates the pixel intensity value FP (= RI + GI + BI) common to each of the R component RI, G component GI, and B component BI of the pixel data, and the correction coefficient calculation unit 13 Although the case where the correction coefficient Ga common to the R component RI, G component GI, and B component BI of the pixel data is calculated has been described, the present invention is not limited to this.

  For example, in the pixel intensity calculation unit 11, using the R component RI, G component GI, and B component BI of the pixel data, three pixel intensity values FP1 (= RI), FP2 (= GI) corresponding to each of them, FP3 (= BI) is obtained, and for each of the three pixel intensity values FP1, FP2, and FP3, the filter unit 12 and the correction coefficient calculation unit 13 perform the same processing as in the case of the pixel intensity value FP. Three correction coefficients Ga1, Ga2, and Ga3 corresponding to the R component RI, G component GI, and B component BI of the data are calculated, and the correction unit 14 calculates the pixel of the target pixel of the sharpened image data Is by Expression (12). Data may be obtained.

  In Expression (12), RO, GO, and BO represent the R component, G component, and B component of the pixel data for the target pixel of the sharpened image data Is. Thereby, sharpened image data Is in which sharpness is enhanced for each of the R component, the G component, and the B component can be obtained in accordance with the degree of blurring of each of the R component, G component, and B component of the pixel data. .

  When the plurality of signal components included in the pixel data constituting the image data I input from the input terminal 15 in FIG. 1 are the Y component, Cb component, and Cr component of the YCbCr signal (each component is YI, In the pixel intensity calculation unit 11, YI can also be used as the pixel intensity value FP. In this case, the filter unit 12 and the correction coefficient calculation unit 13 calculate the correction coefficient Ga of the target pixel P1 for sharpening the image data, and the correction unit 14 calculates the sharpened image data Is according to the equation (13). Pixel data of the target pixel P1 can be obtained.

  In equation (13), YO, CbO, and CrO represent the Y component, Cb component, and Cr component of the pixel data for the target pixel P1 of the sharpened image data Is.

  According to the blur correction processing apparatus 1 configured as described above, the region intensity calculation unit 21 previously provides a plurality of regions Ai having different sizes with the pixel of interest P1 as the center, and for each region Ai, The region intensity value FP is calculated by adding the pixel intensity values of the pixel data in the region Ai, the region selecting unit 22 selects one or more regions from the plurality of regions Ai, and the region intensity filter unit 23 Since the pixel intensity filter value FI of the pixel of interest is calculated from the area intensity value FAi of the selected area Ai, the number of times the filter coefficient is multiplied is reduced, and the amount of calculation of the filter can be reduced while suppressing deterioration in image quality. Therefore, the circuit scale or processing time can be kept small. In addition, with the above configuration, it is possible to obtain filter results with different size regions and different filter coefficients while suppressing the amount of filter calculation.

1 is a block diagram showing a blur correction processing apparatus 1 according to Embodiment 1. FIG. It is a block diagram which shows the filter part 12 of FIG. It is a figure explaining the processing content of the area | region intensity | strength calculation part 21 of FIG. It is the figure which showed an example of several area | region Ai from which the magnitude | size centering on the attention pixel P1 differs. It is the figure which showed an example of the correspondence of blurring degree Bj and several area | region Ai. It is the figure which showed an example of the area | region intensity weighting coefficient ki corresponding to each area | region Ai. It is an example figure which has arrange | positioned the filter coefficient of the pixel in the position of each pixel of area | region Ai.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Blur correction processing apparatus, 11 Pixel intensity | strength calculation part, 12 Filter part, 13 Correction coefficient calculation part, 14 Correction part, 15, 16, 24, 25 Input terminal, 17, 26 Output terminal, 21 Area | region intensity | strength calculation part, 22 area | region Selection unit, 23 area intensity filter unit, Ai area, bst intensity coefficient, Bj blurring degree, FP pixel intensity value, FI pixel intensity filter value, Ga correction coefficient, FAi area intensity value, H area selection information, I image data, Is Sharpened image data, ki, k ₀ i, k ₁ i area intensity weighting coefficients.

Claims (11)

A pixel intensity calculation unit for calculating a pixel intensity value for each pixel data from a plurality of signal components included in each pixel data constituting image data;
A filter unit for calculating a pixel intensity filter value of a target pixel from the pixel intensity value calculated by the pixel intensity calculation unit;
From the pixel intensity value of the target pixel calculated by the pixel intensity calculation unit and the pixel intensity filter value of the target pixel calculated by the filter unit, a correction coefficient of the target pixel for sharpening the image data is calculated. A correction coefficient calculation unit to perform,
A correction unit that sharpens the pixel data of the target pixel of the image data using the correction coefficient of the target pixel calculated by the correction coefficient calculation unit, and obtains pixel data of the sharpened image data;
With
The filter unit is
A plurality of regions having different sizes centered on the target pixel are provided in advance, and the pixel intensity values calculated by the pixel intensity calculation unit are used for each of the regions, and the pixels of the pixel data in the region A region intensity calculator that calculates the region intensity value of each of the regions by adding the intensity values;
An area selection unit for selecting one or more areas from the plurality of areas;
A region intensity filter unit that calculates a pixel intensity filter value of a pixel of interest from the region intensity value calculated by the region intensity calculation unit corresponding to one or more regions selected by the region selection unit;
A blur correction processing apparatus comprising:   The blur correction processing apparatus according to claim 1, wherein the plurality of areas are set so as to include all of the larger areas among those areas.   The region selection unit is configured to select one or more regions from the plurality of regions in accordance with the degree of blurring of the target pixel of the image data based on a correspondence relationship between the plurality of blur levels set in advance and the plurality of regions. The blur correction processing apparatus according to claim 1, wherein the blur correction processing apparatus is selected. The region intensity filter unit includes, for one or more regions selected by the region selection unit, a region intensity value FAi of a region Ai having a region number i (i: integer) and a predetermined region intensity corresponding to the region Ai. 4. The blur according to claim 1, wherein a pixel intensity filter value FI of a target pixel is calculated from the weighting coefficient ki and the number of pixels Ni of the area Ai based on the formula (5). Correction processing device.

The region intensity filter unit includes, for one or more regions selected by the region selection unit, a region intensity value FAi of a region Ai having a region number i (i: integer) and a predetermined region intensity corresponding to the region Ai. The pixel intensity filter value FI of the target pixel is calculated from the weighting coefficient k ₀ i and the number of pixels Ni of the area Ai based on the equation (6). Blur correction processor.

The region intensity filter unit, for one or more regions selected by the region selection unit, the region intensity value FAi of the region Ai with the region number i and a predetermined region intensity weighting coefficient k ₁ i corresponding to the region Ai. The blur correction processing apparatus according to claim 1, wherein the pixel intensity filter value FI of the target pixel is calculated based on the formula (8).

  The pixel intensity calculation unit calculates luminance or a value corresponding to luminance from a plurality of signal components included in pixel data constituting the image data, and outputs the calculated value as the pixel intensity value. The blur correction processing apparatus according to any one of 1 to 6.   The correction coefficient calculation unit calculates the correction coefficient based on a ratio of the pixel intensity filter value of the target pixel calculated by the filter unit to the pixel intensity value of the target pixel calculated by the pixel intensity calculation unit. The blur correction processing apparatus according to claim 1, wherein the blur correction processing apparatus calculates the blur correction processing apparatus according to claim 1.   The correction unit obtains pixel data of the sharpened image data by multiplying pixel data of the target pixel of the image data by the correction coefficient of the target pixel calculated by the correction coefficient calculation unit. The blur correction processing apparatus according to any one of claims 1 to 8,   The blur correction processing apparatus according to claim 1, wherein each of the plurality of regions has a rectangular shape having a different size.   The blur correction processing apparatus according to claim 8, wherein the correction coefficient calculation unit further calculates the correction coefficient based on an enhancement coefficient for adjusting a degree of sharpening of image data.

Images