TWI407801B - Method and apparatus for performing bad pixel compensation - Google Patents

Abstract

A method for performing bad pixel compensation includes: with regard to each direction of a plurality of directions, summing up absolute values of differences between a plurality of sets of first pixel values around a target pixel of an image to generate a first detection value, and with regard to each direction of at least a portion of the directions, summing up absolute values of differences between a plurality of sets of second pixel values around the target pixel to generate a second detection value, where each set of the sets of first pixel values and the sets of second pixel values includes two pixel values corresponding to a difference; and with regard to a specific direction of the directions, selectively performing bad pixel compensation on the target pixel according to the first detection value and the second detection value. An associated apparatus is also provided.

Description

Method for performing bad pixel compensation and means for performing bad pixel compensation

The present invention relates to image processing of Bayer Pattern images, and more particularly to a method and related apparatus for performing bad pixel compensation.

The image produced by the image sensor in a PC Camera is usually a Bayer Pattern image. According to the related art, any sensing unit of the image sensor can only generate pixel values belonging to a color channel among respective pixel values of a corresponding pixel. For example, a complete color channel should contain red, green, and blue color channels, and any sensing unit of the image sensor can only generate one of red, green, and blue pixel values of a corresponding pixel. And the other two pixel values of these red, green, and blue pixel values need to be generated by image processing.

In order to reduce the material cost, the conventional image processing circuit often uses low-order hardware resources, which causes many problems. For example, once the design is too simplistic, there are often side effects that make the image quality poor. In particular, conventional image processing circuits have poor processing effects for bad pixel compensation. Therefore, a novel method is needed to improve image quality.

It is therefore an object of the present invention to provide a method and associated apparatus for performing bad pixel compensation to solve the above problems.

Another object of the present invention is to provide a method and related apparatus for performing bad pixel compensation to ensure image quality. The present invention maintains good image quality even in the case of using low-order hardware resources.

A preferred embodiment of the present invention provides a method for performing bad pixel compensation, the method comprising: calculating, for each direction of a plurality of directions, a first array of values of a complex array near a target pixel in an image The sum of the absolute values of the differences is used as a first detection value, and for each direction of at least a part of the directions, the sum of the absolute values of the differences of the second pixel values of the complex array near the target pixel is calculated as a first And a detection value, wherein each of the first pixel value of the group and the second pixel value of the group includes two pixel values corresponding to a difference, and two pixel values of the same group belong to the same color channel ( a color channel), and each set of first pixel values does not include a pixel value of the target pixel, and each set of second pixel values includes the pixel value of the target pixel; and for a specific direction of the one of the directions, Performing bad pixel compensation on the target pixel selectively according to the first detection value and the second detection value.

The present invention provides a device for performing bad pixel compensation correspondingly, and the device includes: a processing circuit for receiving at least one image signal representing an image, and performing the image on the image Pixel compensation. The processing circuit includes: a detecting module, configured to calculate, in each direction of the plurality of directions, a sum of absolute values of differences of first pixel values of a complex array near a target pixel in an image as a first Detecting a value, and calculating, for each direction of at least a portion of the directions, a sum of absolute values of differences of second pixel values of the complex array near the target pixel as a second detected value, wherein the groups are Each of the one pixel value and the second set of pixel values includes two pixel values corresponding to a difference, two pixel values of the same group belong to the same color channel, and each set of first pixel values does not include a pixel value of the target pixel, and each set of second pixel values includes the pixel value of the target pixel; and a compensation module configured to target the specific one of the directions according to the first detected value The target pixel is selectively subjected to bad pixel compensation with the second detection value.

Please refer to FIG. 1A. FIG. 1A is a schematic diagram of an apparatus 100 for performing bad pixel compensation according to a first embodiment of the present invention. The device 100 includes an image sensor 105 and a processing circuit 110. The processing circuit 110 includes a detection module 112 and a compensation module 114. According to the embodiment, the processing circuit 110 receives at least one image signal S1 representing an image from the image sensor 105, and performs bad pixel compensation on the image, wherein the image is a Bayer Pattern image. As shown in FIG. 1A, the processing circuit 110 can output at least one image signal S2 carrying the compensated image. In addition, the detection module 112 can detect to generate a plurality of detection values, and the compensation module 114 can selectively perform bad pixel compensation on a target pixel according to at least a part of the detection values. In practice, the detection module 112 and the compensation module 114 can be implemented by using a circuit. This is for illustrative purposes only and is not a limitation of the invention. According to some variations of the embodiment, such as the second embodiment shown in FIG. 1B, the processing circuit 110 described above is replaced with the processing circuit 110', and the processing circuit 110' can execute a code 110P for The same operation of the first embodiment, wherein the code 110P includes a program module such as a detection module 112' and a compensation module 114', respectively having the operation of the detection module 112 and the compensation module 114; Variations, the above described device is indicated by the reference numeral 100' in this second embodiment.

Regardless of the architecture shown in FIG. 1A or the architecture shown in FIG. 1B, the present invention can avoid problems of the related art such as poor image quality. In particular, the present invention can properly perform bad pixel compensation. The present invention maintains good image quality even in the case of using low-order hardware resources. Please refer to Figure 2 for details of the operation of the bad pixel compensation.

2 is a flow chart of a method 910 for performing bad pixel compensation in accordance with an embodiment of the present invention. The method 910 can be applied to the device 100 shown in FIG. 1A (or the device 100' shown in FIG. 1B), in particular, the processing circuit 110 of the first embodiment (or the processing circuit 110' of the second embodiment) . For ease of explanation, the following description is made using the architecture of the first embodiment, and the descriptions are also applicable to the architecture of the second embodiment. The method is as follows: In step 912, the detecting module 112 calculates a first pixel value of a complex array near a target pixel in an image (for example, the Bayer pattern image described above) for each of the plurality of directions. The sum of the absolute values of the differences is used as a first detection value, and for each direction of at least a portion of the directions (eg, each of the directions), a complex array second near the target pixel is calculated The sum of the absolute values of the differences in pixel values is used as a second detected value. In particular, each of the set of first pixel values and the set of second pixel values includes two pixel values corresponding to a difference, and the two sets of pixel values of the same group belong to the same color channel. In addition, each set of first pixel values does not include one of the target pixel values, and each set of second pixel values includes the pixel value of the target pixel.

In step 914, the compensation module 114 selectively performs bad pixel compensation on the target pixel according to the first detection value and the second detection value for a specific direction of the directions. In practice, the compensation module 144 can select the specific direction in the directions according to the first detection value in each direction of the directions; and then, the compensation module 114 is configured according to the specific direction. A detected value and the second detected value selectively perform bad pixel compensation on the target pixel. For example, the first detection value in each direction can represent the detection value of the image uniformity in the vicinity of the target pixel, and the compensation module 114 can determine the target pixel according to the size of each first detection value in each direction. The possible edges and/or directions of the neighborhood are nearby as the specific direction described above.

In this embodiment, the directions include N directions such as direction DN(n), where N is a positive integer and n=1, 2, . . . , or N. For example, in the case of N=4, the directions DN(1), DN(2), DN(3), and DN(4) can represent the horizontal direction, the 45-degree direction, the vertical direction, and the 135-degree direction, respectively. For ease of understanding, the four parameters Horz, D45, Vert, and D135 are defined as 1, 2, 3, and 4, respectively, that is, directions DN (1), DN (2), DN (3), and DN. (4) The four index values 1, 2, 3, and 4 are used to indicate the horizontal direction, the 45-degree direction, the vertical direction, and the 135-degree direction in the Pseudo Code in the subsequent description.

3 to 6 illustrate implementation details related to the method 910 shown in FIG. 2 in an embodiment, wherein the circles in FIGS. 3 to 6 represent pixels, and the symbols in the respective circles are, for example, a set. {{V00, V01, V02, V03, V04}, {V10, V11, V12, V13, V14}, {V20, V21, V22, V23, V24}, {V30, V31, V32, V33, V34}, { The symbols in V40, V41, V42, V43, V44}} represent pixel values. Please note that the positive central pixel (the pixel value of V22) in FIGS. 3 to 6 is the target pixel described in step 912. In addition, the arrows in FIGS. 3 to 6 are used to indicate some of the set of pixel values that may be used in step 912, and each arrow corresponds to a set of pixel values, that is, the corresponding step in step 912. Two pixel values of a difference." For example, since the image is a Bayer pattern image in the embodiment, in the case where the pixel value V22 is a blue pixel value, the positions of the arrows in the third to sixth figures are exactly as described in step 912. Description of "two pixel values of the same group belong to the same color channel".

Figure 3 shows the situation in which the direction DN(n) is the horizontal direction DN (Horz) (i.e., in Figure 3, n = Horz), wherein some of the virtual code numbers for the third figure are as follows:

In the above virtual code, the symbol abs in the first few columns represents the absolute value, and linear combinations of these absolute values such as Det1[n] and Det1a[n] (in Figure 3, n=Horz) can be used. The first detected value described in step 912 is generated. For example, in mode M1 (ie, when "mode==1" is established), the first detected value may be the sum of the 12 absolute values of the first six columns of the above virtual code. For another example, in mode M0 (ie, when "mode==0" is established), the first detection value may be 3/2 times the sum of the 8 absolute values of the first 4 columns of the above virtual code, The magnification 3/2 is used to adjust the first detection value, so that the first detection value can be applied to the calculation of the two modes at the same time, so as to reduce the calculation amount and the related cost. Thus, after completing the operation represented by the previous 11 columns in the above virtual code, the symbol Det1[n] (in FIG. 3, n=Horz) may represent the first detection value described in step 912.

Please note that as indicated by the arrow of most of FIG. 3, for each direction DN(n) of the directions described in step 912, such as the horizontal direction DN(1), the set of first pixel values corresponds to the basis. The pixels on at least three straight lines arranged in the direction. For example, in the mode M0, for each direction DN(n) of the directions described in step 912, such as the horizontal direction DN(1), the set of first pixel values correspond to three straight lines arranged according to the direction. Pixel. For another example, in the mode M1, for each direction DN(n) of the directions described in step 912, such as the horizontal direction DN(1), the set of first pixel values correspond to five straight lines arranged according to the direction. The pixels.

In addition, as disclosed by some other arrows in FIG. 3, for each direction DN(n) of the directions described in step 912, such as the horizontal direction DN(1), the set of second pixel values correspond to the basis. The pixels on the straight line arranged in this direction.

In addition, the symbols Det2[n] and Cand[n] (in FIG. 3, n=Horz) in the last two columns of the above virtual code represent the second detection value and the step 914 described in step 912, respectively. The candidate values involved. It is assumed that the specific direction described in step 914 is the direction DN(n0) in the case of n=n0. In step 914, when it is determined that the target pixel is subjected to bad pixel compensation, the compensation module 114 performs bad pixel compensation on the target pixel according to the candidate value Cand[n] (for example, n=n0), wherein the candidate value Cand[n] ] corresponding to one of the set of second pixel values of the particular direction. In particular, the candidate value Cand[n] is a pixel value of the group of second pixel values in the specific direction, such as pixel values {V20, V22} and {V22, V24} except for the pixel value V22 of the target pixel. (In Fig. 3, the average value of the pixel values {V20, V24}). In the case of n0=Horz, this average value is equal to (V20+V24)/2.

Figure 4 shows the situation in which the direction DN(n) is the vertical direction DN(Vert) (i.e., in Fig. 4, n=Vert), and some of the virtual code codes for the fourth figure are as follows:

In the above virtual code, in response to the condition of n=Vert, each pixel value that may be used to calculate the first detection value described in step 912 has been replaced, and each may be used to calculate the number described in step 912. The pixel values of the two detected values have been replaced, and the arrows of Fig. 4 also change positions correspondingly to indicate some sets of pixel values that may be used. In addition, the pixel values required for the calculation of the candidate value Cand[n] (in FIG. 4, n=Vert) have been correspondingly substituted. The similarities between Fig. 4 and Fig. 3 will not be repeated.

Figure 5 shows the direction DN(n) in the 45-degree direction DN (D45) (i.e., in Figure 5, n = D45), wherein some of the virtual code numbers in Figure 5 are detailed below. :

In the above virtual code, in response to the condition of n=D45, each pixel value that may be used to calculate the first detection value described in step 912 has been replaced, and each may be used to calculate the number described in step 912. The pixel values of the two detected values have been replaced, and the arrows of Figure 5 also change positions correspondingly to indicate certain sets of pixel values that may be used. In addition, the pixel values required for the calculation of the candidate value Cand[n] (in FIG. 5, n=D45) have been correspondingly substituted. The similarities between Fig. 5 and Fig. 3 will not be repeated.

Figure 6 shows the direction DN(n) in the 135-degree direction DN (D135) (i.e., in Figure 6, n = D135), wherein some of the virtual code numbers in Figure 6 are detailed below. :

In the above virtual code, in response to the condition of n=D135, each pixel value that may be used to calculate the first detection value described in step 912 has been replaced, and each may be used to calculate the number described in step 912. The pixel values of the two detected values have been replaced, and the arrows of Fig. 6 also change positions correspondingly to indicate some sets of pixel values that may be used. In addition, the pixel values required for the calculation of the candidate value Cand[n] (in FIG. 6, n=D135) have been correspondingly substituted. The similarities between Fig. 6 and Fig. 3 will not be repeated.

According to the embodiment, the de-noising module 114 selects the direction DN(n0) corresponding to the minimum value of the first detection value Det1[n] among the directions DN(n) as the specific direction; that is, The first detected value Det1[n0] is the minimum value among the first detected values Det1[n] in each direction. Thus, if the specific direction in step 914 is the direction DN(n0) in the case of n=n0, the compensation module 114 can use the candidate value Cand[n] under the condition that the condition(s) is satisfied (for example: n=n0) replaces the pixel value of the target pixel, such as the pixel value V22, to perform bad pixel compensation on the target pixel, wherein the virtual code of one of the operations is as follows:

As disclosed in the above virtual code, the compensation module 114 compares the second detection value Det2[n0] of a specific direction DN(n0) with a threshold TH to determine whether to perform bad pixel compensation on the target pixel, wherein the threshold value is TH corresponds to the first detected value Det1[n0] of the specific direction DN(n0). When it is decided to perform bad pixel compensation on the target pixel, the compensation module 114 performs bad pixel compensation on the target pixel. For example, the threshold TH of this embodiment can be described as follows:

The relationship between the threshold TH and the first detected value Det1[n0] of the specific direction DN(n0) is a linear relationship, and the symbols Thrd1 and Thrd2 may represent a fixed value or a predetermined value. In practice, in order to simplify the calculation, the first column in the above virtual code multiplies the second detected value Det2[n0] and the threshold TH by 16 to prevent the compensation module 114 from performing the division operation.

Please note that, according to the embodiment shown in FIG. 3 to FIG. 6, the detecting module 112 calculates, in step 912, the second pixel value of the complex array near the target pixel for each of the directions. The sum of the absolute values of the differences is taken as the second detected value. This is for illustrative purposes only and is not a limitation of the invention. According to some variations of the embodiment, the detecting module 112 calculates, in the step 912, the sum of the absolute values of the differences of the second pixel values of the complex array in the vicinity of the target pixel as the second detection. value. That is, the detection module 112 of these variations does not need to calculate any second detection value for the other directions among the directions (ie, other directions than the specific direction) in step 912.

One of the advantages of the present invention is that the present invention can avoid problems of the related art such as poor image quality. In particular, the present invention can properly perform bad pixel compensation. The present invention maintains good image quality even in the case of using low-order hardware resources.

The above are only the preferred embodiments of the present invention, and all changes and modifications made to the scope of the present invention should be within the scope of the present invention.

100,100’. . . Device for performing bad pixel compensation

105. . . Image sensor

110,110’. . . Processing circuit

110P. . . Code

112, 112’. . . Detection module

114,114’. . . Compensation module

910. . . Method for performing bad pixel compensation

912,914. . . step

S1, S2. . . Image signal

V00, V01, V02, V03, V04, V10, V11, V12, V13, V14, V20, V21, V22, V23, V24, V30, V31, V32, V33, V34, V40, V41, V42, V43, V44. . . Pixel values

1A is a schematic diagram of an apparatus for performing bad pixel compensation in accordance with a first embodiment of the present invention.

1B is a schematic diagram of an apparatus for performing bad pixel compensation in accordance with a second embodiment of the present invention.

2 is a flow chart of a method for performing bad pixel compensation in accordance with an embodiment of the present invention.

3 to 6 illustrate implementation details of the method shown in FIG. 2 in an embodiment.

910. . . Method for performing bad pixel compensation

912,914. . . step

Claims (20)

A method for performing bad pixel compensation, the method comprising: calculating, for each direction of a plurality of directions, a sum of absolute values of differences of first pixel values of a complex array near a target pixel in an image as a first a detection value, and for each direction of at least a part of the directions, calculating a sum of absolute values of differences of second pixel values of the complex array near the target pixel as a second detection value, wherein the groups Each of the first pixel value and the second set of pixel values includes two pixel values corresponding to a difference, and two pixel values of the same group belong to the same color channel (Color Channel), and each group One pixel value does not include one pixel value of the target pixel, and each set of second pixel values includes the pixel value of the target pixel; and for one of the specific directions, according to the first detection value and the The second detected value selectively performs bad pixel compensation on the target pixel. The method of claim 1, wherein the set of first pixel values corresponds to pixels on at least three lines arranged according to the direction for each of the directions. The method of claim 1, wherein the set of second pixel values correspond to pixels on a straight line arranged according to the direction for each of the at least a portion of the directions. The method of claim 1, wherein the step of selectively performing bad pixel compensation on the target pixel according to the first detection value and the second detection value further comprises: according to each of the directions The first detected value of the direction selects the specific direction among the directions. The method of claim 1, wherein the step of selectively performing bad pixel compensation on the target pixel according to the first detection value and the second detection value further comprises: comparing the specific direction a detection value and a threshold value to determine whether to perform bad pixel compensation on the target pixel, wherein the threshold value corresponds to the first detection value in the specific direction; and when it is decided to perform bad pixel compensation on the target pixel, Bad pixel compensation is performed on the target pixel. The method of claim 5, wherein the relationship between the threshold value and the first detected value of the specific direction is a linear relationship. The method of claim 1, wherein the step of selectively performing bad pixel compensation on the target pixel according to the first detection value and the second detection value further comprises: when deciding to perform the target pixel In the case of bad pixel compensation, the target pixel is subjected to bad pixel compensation according to a candidate value, wherein the candidate value corresponds to one of the set of second pixel values in the specific direction. The method of claim 7, wherein the candidate value is an average of pixel values of the set of second pixel values in the specific direction other than the pixel value of the target pixel. The method of claim 7, wherein the step of selectively performing bad pixel compensation on the target pixel according to the first detection value and the second detection value further comprises: replacing the target with the candidate value The pixel value of the pixel to perform bad pixel compensation on the target pixel. The method of claim 1, wherein the image is a Bayer Pattern image. A device for performing bad pixel compensation, the device comprising: a processing circuit for receiving at least one image signal representing an image, and performing bad pixel compensation on the image, wherein the processing circuit comprises: detecting a module, configured to calculate a sum of absolute values of differences of first pixel values of a complex array near a target pixel in an image as a first detection value for each direction of the plurality of directions, and for the directions Calculating, in each direction of at least a part of the target, a sum of absolute values of differences of second pixel values of the complex array in the vicinity of the target pixel as a second detection value, wherein the group of first pixel values and the second group of the groups Each of the pixel values includes two pixel values corresponding to a difference, two pixel values of the same group belong to the same color channel, and each set of first pixel values does not include one of the target pixels a pixel value, and each set of second pixel values includes the pixel value of the target pixel; and a compensation module for using a specific direction of the one of the directions, according to the first detected value and the second detect Measure selection Of the target pixel to compensate for bad pixel. The device of claim 11, wherein for each of the directions, the set of first pixel values correspond to pixels on at least three lines arranged in accordance with the direction. The device of claim 11, wherein the set of second pixel values correspond to pixels on a straight line arranged according to the direction for each of the at least a portion of the directions. The device of claim 11, wherein the compensation module selects the specific direction in the directions according to the first detection value in each direction of the directions. The device of claim 11, wherein the compensation module compares the second detection value and the threshold value in the specific direction to determine whether to perform bad pixel compensation on the target pixel, and the threshold value corresponds to The first detection value of the specific direction; and when determining to perform bad pixel compensation on the target pixel, the compensation module performs bad pixel compensation on the target pixel. The device of claim 15, wherein the relationship between the threshold value and the first detected value of the specific direction is a linear relationship. The device of claim 11, wherein when determining to perform bad pixel compensation on the target pixel, the compensation module performs bad pixel compensation on the target pixel according to a candidate value, and the candidate value corresponds to the specific One of the set of second pixel values of the set of pixel values. The device of claim 17, wherein the candidate value is an average of pixel values of the set of second pixel values in the particular direction other than the pixel value of the target pixel. The device of claim 17, wherein the compensation module replaces the pixel value of the target pixel with the candidate value to perform bad pixel compensation on the target pixel. The device of claim 11, wherein the image is a Bayer Pattern image.