JP2011114683A - Block noise detection device, block noise detection method and block noise detection program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately detect block noise. <P>SOLUTION: A block noise detection device is configured: to decode image data by pixel block, which includes a plurality of pixels composed of at least one image frame arranged in a lattice shape and which is encoded by pixel block; to calculate differences of pixel values between adjacent pixels in the acquired image frame as level differences of pixel boundaries between the pixels; to classify the level difference of each pixel boundary into any of a plurality of groups corresponding to a position of the pixel boundary in each pixel block; to accumulate the classified level difference of each pixel boundary for each classified group; and to specify a block boundary group corresponding to the position to be a boundary between the pixel blocks where pixel boundaries are adjacent to each other, to detect block noise generated in an image frame, according to the value acquired by subtracting the average value of either the accumulated value of the level difference of the pixel boundary in one of groups other than the block boundary groups or the accumulated value of the level difference of the pixel boundary in a plurality of groups from the accumulated value of the level difference of the pixel boundary of the block boundary group. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a block noise detection apparatus for detecting block noise generated when an image signal is encoded and decoded in units of pixel blocks.

As a typical moving picture compression encoding system, for example, there is MPEG-2 (Moving Picture Experts Group-2) which is an international standard. The moving image includes a plurality of images called image frames. MPEG-2 realizes compression of moving images in the time domain by motion compensation in which similar parts are referred to between image frames. In MPEG-2, each image frame is divided into rectangular blocks (hereinafter referred to as pixel blocks) in which a predetermined number of pixels are arranged in a grid in the horizontal and vertical directions, and discrete cosine transform (DCT Discrete Cosine Transform) is performed. , Convert to frequency domain data. The data converted into frequency domain data is quantized and encoded. In MPEG-2, moving picture compression in the spatial domain is realized in this way.

  When DCT is performed, data is biased toward low-frequency components, so that high-frequency components may be lost due to quantization. For example, when encoding is performed at a high compression rate, high-frequency components are easily lost due to quantization. Due to the lack of high frequency components, the continuity of pixel values near the boundary between adjacent pixel blocks is lost in the decoded image frame, and rectangular noise is generated in the screen. This rectangular noise is called block noise.

  For example, a difference between pixel values of adjacent pixels is used as a determination value for determining the occurrence state of block noise.

JP 2005-12461 A JP 2000-350202 A JP 2000-50275 A JP 2002-204457 A

  However, the difference in pixel value between adjacent pixels includes a difference due to the pattern of the image in addition to a difference due to block noise. Therefore, when the difference between the pixel values between adjacent pixels is used as the determination value for the block noise occurrence state, the difference between the pixel values due to the pattern is included in the determination value, so that the detection result of the block noise occurrence state is included in the pattern. There was a problem of being dependent. For example, when the image pattern is flat such that it is a single color on one side, the change in the pixel value is small, but in an image with a complicated pattern, the change in the pixel value is intense. Therefore, when the degree of image quality degradation due to block noise of the image is about the same, the more complex the image, the greater the difference in the pixel value of the image, and the greater the dependency of the detection result on the occurrence of block noise on the image. .

  An object of one aspect of the present invention is to provide a block noise detection apparatus that can accurately detect a block noise occurrence state.

One aspect of the present invention is a block noise detection apparatus. This block detector is
A target image frame obtained by receiving image data encoded in units of pixel blocks each including a plurality of pixels in which at least one input image frame is arranged in a grid, and decoding each of the image data in units of pixel blocks In the block noise detection device for detecting block noise in
A level difference calculating unit that calculates a difference in pixel value between adjacent pixels in the target image frame as a pixel boundary level of a pixel boundary between the pixels;
The calculated pixel boundary step is classified into one of a plurality of groups corresponding to the position of the pixel boundary in each pixel block, and the pixel boundary step is accumulated for each classified group. And
Among the plurality of groups, a block boundary group corresponding to a position where a pixel boundary is a boundary between adjacent pixel blocks is specified, and a group other than the block boundary group is determined from a cumulative value of pixel boundary steps of the block boundary group A value obtained by subtracting an accumulated value of pixel boundary steps of one group of the above or an average value of accumulated values of pixel boundary steps of a plurality of groups other than the block boundary group as a block noise determination value And
A detection unit that detects a block noise occurrence state in the target image frame according to the block noise determination value.

  Another aspect of the present invention is the block noise detection method described above. Another aspect of the present invention can include a program that causes a computer to function as an information processing apparatus, and a computer-readable recording medium that records the program.

  According to the disclosed block noise detection apparatus, it is possible to accurately detect a block noise occurrence state.

It is a figure which shows the structural example of a block noise detection system. It is a figure which shows an example of an image frame. It is a figure for demonstrating the example which estimates the pixel value in a pixel boundary by high-order prediction. It is a figure for demonstrating the example which estimates the pixel value in a pixel boundary by high-order prediction. It is a figure for demonstrating the example which estimates the pixel value in a pixel boundary by linear prediction. It is a figure for demonstrating the example which estimates the pixel value in a pixel boundary by linear prediction. It is a figure showing the case where each pixel value of a plurality of arranged pixels changes with a substantially constant gradient. It is an example showing a comparison between a difference absolute value A of pixel values of a pixel x4 and a pixel x5, a difference absolute value B of an estimated pixel value y _P4 and an estimated pixel value y _P5 , and a predetermined threshold value T1. It is an example showing a comparison between a difference absolute value A of pixel values of a pixel x4 and a pixel x5, a difference absolute value B of an estimated pixel value y _P4 and an estimated pixel value y _P5 , and a predetermined threshold value T1. It is a figure which shows the example which accumulate | stores the pixel boundary level for one image frame for every period of the size of a pixel block. It is a figure which shows the result of accumulation | storage of the pixel boundary level | step difference when the size of a pixel block is assumed to be 6 pixels in a horizontal direction and 6 pixels in a vertical direction. It is a figure which shows the result of accumulation | storage of the pixel boundary level | step difference when the size of a pixel block is assumed to be 8 pixels in a horizontal direction and 8 pixels in a vertical direction. It is a figure which shows the result of accumulation | storage of the pixel boundary level | step difference when the size of a pixel block is assumed to be 16 pixels in a horizontal direction, and 16 pixels in a vertical direction. It is a figure which shows the comparison with the cumulative value of the pixel boundary level | step difference in a block boundary, and the cumulative value of the pixel boundary level | step difference in a non-block boundary. It is a figure which shows the flow of a detection process of the block noise generation strength of a block noise detection apparatus. It is a figure which shows the structural example of a block noise detection system. It is a figure for demonstrating the detection process of an accumulation object area | region. It is a figure which shows the structural example of a block noise reduction system.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The configuration of the following embodiment is an exemplification, and the present invention is not limited to the configuration of the embodiment.

<First Embodiment>
In the first embodiment, the block noise detection apparatus detects a block noise occurrence state in each image frame included in an image frame group obtained by decoding an encoded moving image signal.

  FIG. 1 is a diagram illustrating a configuration example of a block noise detection system. The block noise detection system 100A includes an input unit 11, a decoding unit 12, a block noise detection device 1A, a network 5, a storage device 6, and a portable recording medium 7. The block noise detection device 1 </ b> A includes a pixel boundary step calculation unit 13, a pixel boundary step accumulation unit 14, a determination value calculation unit 15, and a block noise determination unit 16.

  The input unit 11 obtains, for example, a moving image signal compressed and encoded by MPEG-2 as an input. The compressed and encoded video signal is compressed and encoded by, for example, a moving image distribution server (not shown) connected to the network 5, sent to the network 5, and input to the input unit 11 via the network 5. Is done. Alternatively, the compression-coded moving image signal is stored in the storage device 6 provided inside or outside the block noise detection device 1 </ b> A, read out from the storage device 6, and input to the input unit 11. The storage device 6 is, for example, a hard disk drive (HDD). Alternatively, the moving image signal is recorded on the portable recording medium 7 and is read from the portable recording medium 7 by the portable recording medium driving device provided inside or outside the block noise detection device 1A and input to the input unit. 11 is input. The portable recording medium 7 is, for example, a DVD, a Blu-ray disc, a USB memory, or the like. The input unit 11 outputs the compressed and encoded moving image signal to the decoding unit 12.

  The decoding unit 12 obtains the moving image signals that have been compression-encoded from the input unit 11 as sequential inputs. The decoding unit 12 decodes the compressed and encoded moving image signal to generate an image frame group.

  FIG. 2 is a diagram illustrating an example of an image frame. FIG. 2 shows an image frame F1 and an enlarged view of a pixel block B1, which is one of the pixel blocks included in the image frame F1. The moving image signal is encoded and decoded in units of a rectangular pixel block including a predetermined number of pixels in the horizontal direction and the vertical direction. For example, MPEG-2 uses a pixel block of 8 pixels in the horizontal direction and 8 pixels in the vertical direction. FIG. 2 shows an image frame obtained by decoding a moving image signal compression-encoded by MPEG-2. The pixel block has 8 pixels in the horizontal direction (row) and 8 pixels in the vertical direction (column). The size of the pixel.

  In the first embodiment, a boundary between adjacent pixels is referred to as a pixel boundary. A pixel boundary located at a boundary between adjacent pixel blocks is referred to as a block boundary. Pixel boundaries other than block boundaries are referred to as non-block boundaries.

  In the pixel block B1 in the example shown in FIG. 2, pixels are indicated by squares. A boundary between pixels adjacent to each other in the horizontal direction is a pixel boundary in the horizontal direction. A boundary between pixels adjacent in the vertical direction is a pixel boundary in the vertical direction. Each pixel boundary located on the vertical line indicated by reference characters a7 and b7 and each pixel boundary located on the horizontal line indicated by reference characters c7 and d7 are block boundaries. Each pixel boundary located on the vertical line indicated by reference signs b0 to b6 and a non-block boundary in the horizontal direction located on the horizontal line indicated by reference signs d0 to d6.

  In the image frame F1 of the example shown in FIG. 2, the X axis and the Y axis are set in the horizontal direction with the pixel p0 located at one end of the top row of the pixels included in the image frame F1 as the origin. The coordinates of the pixel p0 are expressed as (X, Y) = (0, 0).

  In the first embodiment, the decoding unit 12 outputs the pixels included in each image frame included in the decoded image frame group to the block noise detection device 1A. The decoding unit 12 outputs the pixels included in each image frame in the order in which the image frames are to be reproduced. The decoding unit 12 outputs the pixels included in each image frame in the order of the + X-axis direction, starting from the pixel p0 located at the origin and the pixel located on Y = 0. When the decoding unit 12 outputs a pixel located at the other end of the pixel p0 of the pixel group aligned on Y = 0, the pixel located at the coordinate (X, Y) = (0, 1) is next. From, the pixels located on Y = 1 are output in the order of the + X-axis direction. When the pixel located at the end of the pixel group arranged on Y = 1 is output, the pixel located at the coordinates (X, Y) = (0, 2) is output next. In this way, the pixels included in the image frame are output to the block noise detection device 1A.

  Pixels are input from the decoding unit 12 to the block noise detection device 1A. The block noise detection apparatus 1A is an apparatus that detects a block noise occurrence state of an image frame formed from an input pixel group.

  Pixels are sequentially input to the pixel boundary step calculation unit 13 in the order output from the decoding unit 12 for each image frame included in the image frame group. The pixel boundary step calculation unit 13 obtains a difference in pixel value at a pixel boundary between pixels adjacent in the horizontal direction included in each image frame from sequentially input pixels. This pixel value step (difference) at the pixel boundary is hereinafter referred to as a pixel boundary step. The pixel value is a value of information for expressing one pixel. When the image is a monochrome image, one pixel is represented by a luminance value. When the image is a color image, for example, one pixel is represented by three pieces of information of a luminance value and two color difference components (a difference between blue and luminance values and a difference between red and luminance values). In the first embodiment, regardless of whether the image is a monochrome image or a color image, the process using the pixel value is performed using the luminance value.

  The pixel boundary step calculation unit 13 calculates the absolute difference value of the pixel values between adjacent pixels. In addition, the pixel boundary step calculation unit 13 estimates the pixel value at the pixel boundary from each side using the pixel values of a plurality of pixels located on both sides of the pixel boundary by high-order prediction or linear prediction, and the like. Calculate an estimate. The pixel boundary step calculation unit 13 calculates the absolute difference between the two estimated values. The pixel boundary step calculation unit 13 determines the smaller one of the absolute difference value of the pixel values between adjacent pixels and the absolute difference value of the two estimated values at the pixel boundary as the pixel boundary step.

  3A and 3B are diagrams for explaining an example of estimating a pixel value at a pixel boundary by high-order prediction. 3A and 3B show graphs of pixel values of the pixels x1 to x8 arranged in order in the horizontal direction. The pixel boundary step calculation unit 13 obtains a pixel boundary step at the pixel boundary P4 between the pixel x4 and the pixel x5.

In the example illustrated in FIGS. 3A and 3B, the pixel boundary step calculation unit 13 may, for example,
The pixel value at the pixel boundary P4 is estimated from the pixel x4 side by high-order prediction using the three pixels x2, x3, and x4 located on the fourth side. The pixel value estimated from the pixel x4 side of the pixel boundary P4 is referred to as an estimated pixel value _yp4 . Similarly, the pixel boundary step calculation unit 13 estimates the pixel value at the pixel boundary P4 from the pixel x5 side by high-order prediction using, for example, three pixels x5, x6, and x7 located on the pixel x5 side. To do. The pixel value estimated from the pixel x5 side of the pixel boundary P4 is referred to as an estimated pixel value _yp5 .

In the example shown in FIGS. 3A and 3B, the pixel boundary step calculation unit 13 uses a curve y = ax ² representing a pixel value change from the pixel x2, the pixel x3, the pixel x4, and the respective pixel values y2, y3, y4. + Bx + c is estimated. A higher order prediction of the curve y = ax ² + bx + c is shown below.

Since the difference between the estimated pixel value obtained from the estimated curve y = ax ² + bx + c and each pixel value is minimized, the pixel boundary step calculation unit 13 firstly determines each pixel value and the estimated curve y = ax ² + bx + c.
The sum of squares of the difference from the estimated pixel value obtained from is obtained.

  When the pixel value in the pixel xi (i is an integer of 0 or more) is yi, the sum of squares of the difference between the estimated pixel value and the pixel value is expressed by the following Expression 1.

推定画素値と画素値との差分が最小となるためには、

In order to minimize the difference between the estimated pixel value and the pixel value,

である。したがって、

It is. Therefore,

である。よって、

It is. Therefore,

となる。これを行列式で表すと、以下の式２になる。

It becomes. When this is expressed by a determinant, the following expression 2 is obtained.

ここで、

here,

とおくと、式２は、

Then, Equation 2 is

と表わされる。従って、

It is expressed as Therefore,

となり、推定曲線（画素値変化を表す曲線）ｙ＝ａｘ²＋ｂｘ＋ｃの係数ａ，ｂ，ｃを算
出することができる。

Thus, it is possible to calculate the coefficients a, b, and c of the estimated curve (curve representing pixel value change) y = ax ² + bx + c.

The pixel boundary step calculation unit 13 obtains an estimated pixel value y _P4 on the pixel x4 side of the pixel boundary P4 using an estimated curve calculated from the pixels x2, x3, x4 and the respective pixel values y2, y3, y4. . Similarly, the pixel boundary step calculation unit 13 calculates an estimation curve from the pixels x5, x6, and x7 and the respective pixel values y5, y6, and y7, and calculates an estimated pixel value y _p5 on the pixel x5 side of the pixel boundary P4. Ask. For the x coordinate of the pixel boundary P4, for example, the estimated pixel values y _P4 and y _P5 of the pixel boundary P4 may be obtained using x _P4 = (x4 + x5) / 2 which is intermediate between the pixel x4 and the pixel x5. .

The pixel boundary step calculation unit 13 is the smaller of the difference absolute value A between the pixel value of the pixel x4 and the pixel value of the pixel x5 and the difference absolute value B between the estimated pixel value _yp4 and the estimated pixel value _yp5. Is determined as a pixel boundary step at the pixel boundary P4.

In the example shown in FIG. 3A, the difference absolute value A between the pixel value of the pixel x4 and the pixel value of the pixel x5 is smaller than the difference absolute value B between the estimated pixel value y _P4 and the estimated pixel value y _P5. A is adopted as the pixel boundary step at the pixel boundary P4.

In the example shown in FIG. 3B, the difference absolute value B between the estimated pixel value _yp4 and the estimated pixel value _yp5 is smaller than the difference absolute value A between the pixel value of the pixel x4 and the pixel value of the pixel x5. B is adopted as the pixel boundary step at the pixel boundary P4.

  4A and 4B are diagrams for explaining an example of processing for obtaining a pixel boundary step by linear prediction. 4A and 4B, similarly to FIGS. 3A and 3B, are graphs of the pixel values of the pixel x1 to the pixel x8, and the pixel boundary step at the pixel boundary P4 between the pixel x4 and the pixel x5 is obtained. Shown about the case.

In the example illustrated in FIGS. 4A and 4B, the pixel boundary step calculation unit 13 uses the pixels y3 and x4 on the pixel x4 side of the pixel boundary P4 and the pixel values y3 and y4 to represent a straight line y representing a pixel value change. = Ax + b is estimated, and an estimated pixel value _yp4 at the pixel boundary P4 is obtained. Similarly, the pixel boundary step calculation unit 13 estimates a straight line representing a change in pixel value from the pixels x5 and x6 on the pixel x5 side of the pixel boundary P and the respective pixel values y5 and y6, and at the pixel boundary P4. An estimated pixel value y _P5 is obtained. Since the linear coefficients a and b representing the pixel value change can be easily obtained by solving the simultaneous linear equations, description of the straight line estimation method and the estimated pixel value representing the pixel value change is omitted.

In the example shown in FIG. 4A, the difference absolute value A between the pixel value of the pixel x4 and the pixel value of the pixel x5 is smaller than the difference absolute value B between the estimated pixel value y _P4 and the estimated pixel value y _P5. A is adopted as the pixel boundary step at the pixel boundary P4.

In the example shown in FIG. 4B, the difference absolute value B between the estimated pixel value _yp4 and the estimated pixel value _yp5 is smaller than the difference absolute value A between the pixel value of the pixel x4 and the pixel value of the pixel x5. B is adopted as the pixel boundary step at the pixel boundary P4.

By considering the estimated pixel value at the pixel boundary in determining the pixel boundary step, it is possible to more accurately obtain the difference in the pixel value between adjacent pixels depending on the design. For example, when the pixel values of a plurality of arranged pixels change with a substantially constant gradient, the pixel boundary step can be obtained more accurately.

  FIG. 5 is a diagram illustrating a case where the pixel values of a plurality of arranged pixels change with a substantially constant gradient. When the pixel values between the pixels are continuous as in the example shown in FIG. 5, it is considered that there is almost no step due to the pattern at each pixel boundary. Further, in the pixel boundary P4 in the example shown in FIG. 5, the absolute difference value of the estimated pixel value is smaller than the absolute difference value of the pixel values of the pixel x4 and the pixel x5. Therefore, at the pixel boundary where the pixel values between the pixels are continuous, the difference absolute value of the estimated pixel value smaller than the difference absolute value of the pixel values of the pixel x4 and the pixel x5 is determined as the pixel boundary step. It is possible to obtain the pixel boundary step at the pixel boundary considered not to be more accurate.

  The pixel boundary step calculation unit 13 outputs the pixel boundary step obtained using the higher-order prediction or linear prediction described in FIGS. 3A, 3B, 4A, and 4B to the pixel boundary step accumulation unit 14. .

  The pixel boundary step accumulation unit 14 receives the pixel boundary step from the pixel boundary step calculation unit 13 as an input. The pixel boundary step accumulation unit 14 classifies the input pixel boundary step into one group corresponding to the position of the pixel boundary in the pixel block in the group having the same number as the size of the pixel block. The pixel boundary step accumulation unit 14 accumulates pixel boundary steps for each classified group and detects a block boundary group from the accumulation result of each group. In addition, when the pixel block size is unknown, the pixel boundary step accumulation unit 14 classifies the pixel boundary step into each group by using a plurality of pixel block size candidate values held in advance. Accumulate and detect pixel block size and block boundary.

  When the pixel boundary step accumulation unit 14 cumulatively calculates the pixel boundary step, if the pixel boundary step is greater than or equal to a predetermined threshold T1, it is considered that the pixel boundary step has become large because the step due to the pattern is large. Therefore, it is excluded from the cumulative calculation. The predetermined threshold value T1 is a value that is sufficiently larger than a step that is considered to be caused by block noise, and is a threshold value for determining whether or not to add a pixel boundary step at each pixel boundary to the cumulative calculation.

  The pixel boundary step accumulation unit 14 determines whether the pixel boundary step is smaller than a predetermined threshold T1. The pixel boundary step accumulation unit 14 adds the pixel boundary step to the cumulative calculation when the pixel boundary step is smaller than the predetermined threshold T1, and excludes the pixel boundary step from the cumulative calculation when the pixel boundary step is equal to or larger than the predetermined threshold T1.

6A and 6B show a difference absolute value A between the pixel value of the pixel x4 and the pixel value of the pixel x5, a difference absolute value B between the estimated pixel value y _P4 and the estimated pixel value y _P5 of the pixel boundary P4, and a predetermined value. It is an example which shows a comparison with threshold value T1.

  In the example shown in FIG. 6A, since the difference absolute value A is smaller than the difference absolute value B, the difference absolute value A is determined as the pixel boundary step. Since the difference absolute value A is smaller than the predetermined threshold T1, the pixel boundary step accumulation unit 14 accumulates the difference absolute value A as a pixel boundary step.

  In the example shown in FIG. 6B, since the difference absolute value B is smaller than the difference absolute value A, the difference absolute value B is determined as the pixel boundary step. Since the difference absolute value B is larger than the predetermined threshold value T1, the pixel boundary step accumulation unit 14 excludes the difference absolute value B as the pixel boundary step from the accumulation calculation.

The pixel boundary step accumulation unit 14 accumulates pixel boundary steps determined to be added to the accumulation calculation by comparison with the threshold value T1 among pixel boundary steps for one image frame.

  FIG. 7 is a diagram illustrating an example in which pixel boundary steps for one image frame are accumulated for each pixel block size period (block period).

  A pixel boundary between a pixel xi (i is an integer of 0 or more) adjacent in the horizontal direction and a pixel xi + 1 is Pi, and a column size (block period) of the pixel block is S. i is a variable indicating the order in which pixels are input to the pixel boundary step accumulation unit 14.

  The pixel boundary step accumulation unit 14 uses the variable n (n: an integer equal to or greater than 0) to set the value of the remainder m when the variable i = S × n + m (m: 0 ≦ m <an integer of S). Accordingly, the pixel boundaries Pi are classified into groups. The pixel boundary step accumulation unit 14 accumulates pixel boundary steps in the classified groups.

  For example, when the moving image is compression-encoded by MPEG-2, since the size (block period) of the pixel block is 8, S = 8. In the pixel boundary step accumulation unit 14, the remainder m when i = 8 × n + m is m = 0, m = 1, m = 2, m = 3, m = 4, m = 5, m = 6, m The pixel boundary step of the pixel boundary Pi is classified into the group of the value of the remainder m of the variable i in the group of = 7. The pixel boundary step accumulation unit 14 accumulates pixel boundary steps in each classified group. Finally, the remainder m when i = 8 × n + m is m = 0, m = 1, m = 2, m = 3, m = 4, m = 5, m = 6, and m = 7. Pixel boundary steps are accumulated for each group of pixel boundaries Pi. That is, the same number (eight) cumulative values as the pixel block column size are obtained.

  When the pixel boundary steps of the pixel boundary Pi are classified into groups according to the remainder m of the variable i when i = S × n + m, for example, the pixels in the example shown in FIG. The pixel boundary step of the pixel boundary in each pixel block corresponding to each pixel boundary located on each vertical line indicated by the symbols b0 to b7 of the block B1 is classified. For example, the group of remainder m = 5 is a group of pixel boundary steps at the pixel boundary corresponding to each pixel boundary located on the vertical line indicated by the symbol b0 of the pixel block B1 in FIG. 2 in each pixel block in the image frame. It is. When pixel boundary steps are classified in this way, the pixel boundary steps at each pixel boundary that becomes a block boundary are classified into one group.

  When an image is encoded with a high compression rate in pixel block units, high-frequency components are missing, and continuity of low-frequency components is guaranteed within pixel blocks, but not missing between pixel blocks. The continuity of low frequency components is not guaranteed. For this reason, pixel values are discontinuous between pixel blocks. This discontinuity of low frequency components causes a step between pixel blocks due to block noise. Therefore, at the block boundary where a step due to block noise occurs, the pixel boundary step becomes larger than the pixel boundary step at the non-block boundary. That is, the cumulative value of the pixel boundary step of the group in which the pixel boundary step of the pixel boundary serving as the block boundary is the largest among all the groups.

  In the example shown in FIG. 7, since the cumulative value of the group of pixel boundary steps at the pixel boundary Pi where the remainder m when m = 3 is expressed as i = 8 × n + m is the maximum, The unit 14 detects that the pixel boundary Pi where i = 8 × n + 3 is a block boundary. In the example shown in FIG. 7, a group with a remainder m = 3 is a block boundary group, and each group with a remainder m = 0, 1, 2, 4, 5, 6, 7 is a non-block boundary group.

When the size of the pixel block is unknown, the pixel boundary step accumulation unit 14 uses a plurality of preset block period candidate values one by one, and sets the pixel boundary according to the position of the pixel boundary in the pixel block. The cumulative value of the pixel boundary step of each group in which the step is classified is obtained. Based on the accumulated value of each group obtained by accumulating the pixel boundary step using the candidate value of each block period, the pixel boundary step accumulation unit 14 detects the size of the pixel block and the pixel boundary that becomes the block boundary. To do.

  8A, 8B, and 8C are diagrams illustrating examples of results obtained by accumulating pixel boundary steps in an image frame using candidate values having different block periods. FIG. 8A shows the result of accumulating pixel boundary steps when 6 is used as a candidate value for the block period, that is, when the size of the pixel block is assumed to be 6 pixels in the horizontal direction and 6 pixels in the vertical direction. FIG. 8B shows the result of accumulation of pixel boundary steps when 8 is used as a candidate value for the block period, that is, when the size of the pixel block is assumed to be 8 pixels in the horizontal direction and 8 pixels in the vertical direction. FIG. 8C shows the result of accumulating pixel boundary steps when 16 is used as a candidate value for the block period, that is, when the size of the pixel block is assumed to be 16 pixels in the horizontal direction and 16 pixels in the vertical direction.

  The pixel boundary step accumulation unit 14 accumulates pixel boundary steps included in the image frame by using a plurality of block period candidate values one by one, for example, the example shown in FIGS. 8A, 8B, and 8C. The cumulative result like The pixel boundary step accumulation unit 14 analyzes the obtained accumulation result and detects the size of the pixel block.

  When the pixel block size is appropriate, the pixel boundary step at the pixel boundary that becomes the block boundary is concentrated and classified in one group, so that the cumulative value of the pixel boundary step in the one group is the value of another group. This is a prominent value compared to the accumulated value of the pixel boundary step. On the other hand, when the block size is not appropriate, the pixel boundary step at the pixel boundary that is the block boundary is distributed and classified, and therefore, the group in which the cumulative value of the pixel boundary step protrudes compared to other groups. Does not appear.

  The pixel boundary step accumulation unit 14 calculates the smallest difference among the accumulated value differences between the largest accumulated value group and other groups in the accumulated result obtained using the candidate values of the respective block periods. To detect. That is, the pixel boundary step accumulation unit 14 calculates a difference between accumulated values of the group having the largest accumulated value and the group having the next largest accumulated value from the accumulated results obtained by using the candidate values of the respective block periods. Ask.

  The pixel boundary step accumulation unit 14 detects the candidate value of the block period when the difference between the accumulated values of the group with the largest accumulated value and the group with the next largest accumulated value becomes the largest as the size of the pixel block.

  In the example of the accumulation result when the block period is assumed to be 6 shown in FIG. 8A, the accumulated value of the pixel boundary step of the group in which the remainder m = 4 when i = 6 × n + m is expressed is the maximum. It is. Next, the accumulated value of the pixel boundary step of the group in which the remainder m = 3 is large.

  In the example of the accumulation result when the block period is assumed to be 8 shown in FIG. 8B, the accumulated value of the pixel boundary step of the group of the remainder m = 3 is the maximum when i = 8 × n + m. . Next, the cumulative value of the pixel boundary step of the group in which the remainder m = 6 is large.

  In the example of the accumulation result when the block period is assumed to be 16 shown in FIG. 8C, the accumulated value of the pixel boundary step of the group of the remainder m = 3 is the maximum when i = 8 × n + m. . Next, the cumulative value of the pixel boundary step of the group in which the remainder m = 11 is large.

8A, 8B, and 8C, the difference in the accumulated value between the group with the largest accumulated value and the group with the next largest accumulated value is the largest in the example shown in FIG. 8B. Accordingly, the pixel boundary step accumulation unit 14 detects that the pixel block size is 8, and that the block boundary is a pixel boundary Pi where i = 8 × n + 3.

  The pixel boundary step accumulation unit 14 outputs the detected pixel block size, block boundary, non-block boundary, and the accumulated value of the pixel boundary step of each group to the determination value calculation unit 15.

  The determination value calculation unit 15 receives, from the pixel boundary step accumulation unit 14, the detected pixel block size, block boundary, non-block boundary, and cumulative values of pixel boundary steps of each group as inputs. The determination value calculation unit 15 calculates a block noise detection determination value by subtracting the average of the pixel boundary step of the non-block boundary group from the cumulative value of the pixel boundary step of the block boundary group.

  When the accumulation result illustrated in FIG. 8B is input, the determination value calculation unit 15 determines the pixel boundary step on the non-block boundary from the accumulation value of the group of m = 3 that is the cumulative value of the pixel boundary step on the block boundary. The block noise detection determination value is calculated by subtracting the average of the accumulated values of the seven groups m = 0, 1, 2, 4, 5, 6, and 7.

  FIG. 9 is a diagram showing a comparison between the accumulated value of the pixel boundary step of the block boundary group and the accumulated value of the pixel boundary step of a certain non-block boundary group. Since a pixel block is often small enough for the size of a certain pattern area in an image frame, the level difference due to the pattern at the block boundary is almost equal to the level difference due to the pattern at the non-block boundary in the entire image frame. Can be considered. Further, it can be considered that the pixel boundary step at the block boundary includes a portion corresponding to the step due to the pattern and a portion corresponding to the step due to the block noise. On the other hand, the pixel boundary step at the non-block boundary is considered to be a step due to the pattern.

  Therefore, the value obtained by subtracting the accumulated value of the pixel boundary step of a certain non-block boundary group from the accumulated value of the pixel boundary step of the block boundary group is considered to be a value that approximates the accumulated value of the step due to block noise in the image frame. It is done. In the first embodiment, the determination value calculation unit 15 subtracts the average value of the accumulated pixel boundary steps of each non-block boundary group from the accumulated value of the pixel boundary steps of the block boundary group as the block noise detection determination value. In this way, a value that approximates the cumulative value of the step due to block noise in the image frame is calculated as the block noise detection determination value.

  The determination value calculation unit 15 outputs the calculated block noise detection determination value of the processing target image frame to the block noise determination unit 16.

  The block noise determination unit 16 receives the block noise detection determination value of the image frame to be processed from the determination value calculation unit 15 as an input. The block noise determination unit 16 holds a plurality of block noise generation intensity threshold values in advance, compares the block noise detection determination value with the block noise generation intensity threshold value, and determines the image quality of the image frame to be processed by block noise. The block noise generation intensity indicating the degree of influence is determined.

  For example, the block noise determination unit 16 holds a block noise generation intensity threshold 1, a block noise generation intensity threshold 2, and a block noise generation intensity threshold 3. Block noise generation intensity threshold 1 <block noise generation intensity threshold 2 <block noise generation intensity threshold 3.

When the block noise detection determination value is less than the block noise occurrence intensity threshold value 1, the block noise determination unit 16 is such that image quality deterioration due to block noise is negligible for the human eye, and thus the image frame to be processed It is determined that no block noise occurs in

  When the block noise detection determination value is not less than the block noise occurrence intensity threshold 1 and less than the block noise occurrence intensity threshold 2, the block noise determination unit 16 determines that the block noise occurrence intensity in the processing target image frame is weak.

  When the block noise detection determination value is greater than or equal to the block noise occurrence intensity threshold 2 and less than the block noise occurrence intensity threshold 3, the block noise determination unit 16 determines that the block noise occurrence intensity in the processing target image frame is medium.

  When the block noise detection determination value is equal to or greater than the block noise occurrence intensity threshold 3, the block noise determination unit 16 determines that the block noise occurrence intensity in the processing target image frame is strong. The block noise determination unit 16 outputs the block noise occurrence intensity.

  By detecting the block noise occurrence intensity in the processing target image frame, the occurrence state of the block noise in the processing target image frame corresponding to the block noise generation intensity is detected.

  The pixel boundary step calculation unit 13 corresponds to a step calculation unit. The pixel boundary step accumulation unit 14 corresponds to an accumulation unit. The determination value calculation unit 15 corresponds to a calculation unit. The block noise determination unit 16 corresponds to a detection unit.

  FIG. 10 is a diagram showing a flow of block noise generation intensity detection processing of the block noise detection apparatus 1A. The pixel boundary level calculation unit 13 starts processing when a predetermined number of pixels are buffered after the decoding unit 12 starts to input pixels of the image frame to be processed.

The pixel boundary step calculation unit 13 performs estimated pixel values y _Pi and y _{Pi + at} the pixel boundary Pi between the pixel Xi and the pixel Xi + 1 that are adjacent in the horizontal direction by high-order prediction or linear prediction for sequentially input pixels. ₁ is calculated (OP1). In addition, for example, when obtaining an estimated pixel value at the pixel boundary P1 between the pixel X1 and the pixel X2, or when obtaining an estimated pixel value at the last pixel boundary Pi of the image frame, If there is no necessary data, OP1 is omitted.

The pixel boundary step calculation unit 13 determines whether or not the absolute difference value between the pixel values Xi and Xi + 1 is smaller than the absolute difference value between the estimated pixel values y _Pi and y _{Pi + 1} between the pixel Xi and the pixel Xi + 1. Is determined (OP2). When the difference absolute value of the pixel values between the pixel Xi and the pixel Xi + 1 is smaller than the difference absolute value between the estimated pixel values y _Pi and y _{Pi + 1} between the pixel Xi and the pixel Xi + 1 (OP2: Yes), The pixel boundary step calculation unit 13 sets the absolute difference between the pixel values of the pixel Xi and the pixel Xi + 1 as the pixel boundary step of the pixel boundary Pi (OP3). When the difference absolute value of the estimated pixel values of the pixel Xi and the pixel Xi + 1 is smaller than the difference absolute value of the pixel values of the pixel Xi and the pixel Xi + 1 (OP2: No), the pixel boundary step calculation unit 13 The absolute difference between the estimated pixel values of the pixel Xi and the pixel Xi + 1 is set as a pixel boundary step of the pixel boundary Pi (OP4). Further, for example, when there is no data necessary for obtaining the estimated pixel value as in the case of obtaining the estimated pixel value at the pixel boundary P1 between the pixel X1 and the pixel X2, the pixel values of the pixel X1 and the pixel X2 The absolute difference value is determined as the pixel boundary step. The pixel boundary step calculation unit 13 outputs the pixel boundary step of the pixel boundary Pi to the pixel boundary step accumulation unit 14.

The pixel boundary step accumulation unit 14 determines whether or not the pixel boundary step at the pixel boundary Pi is smaller than the threshold T1 (OP5). When the pixel boundary step is smaller than the threshold T1 (OP5: Yes)
The pixel boundary step accumulation unit 14 accumulates the pixel boundary step in the group indicated by the remainder m when the variable i is expressed as i = S × n + m (OP6). At this time, if the size of the pixel block is not known, the pixel boundary step accumulation unit 14 uses each of the candidate values of the plurality of block periods, and uses the pixel boundary in the image frame for each candidate value of the block period. The accumulated value of the step is obtained, and the size of the pixel block of the moving image signal is detected from the accumulated result. If the pixel boundary step is greater than or equal to the threshold T1 (OP5: No), the process returns to OP1.

  The pixel boundary step accumulation unit 14 determines whether the current pixel boundary Pi to be processed is a pixel boundary before the last pixel boundary included in the image frame (OP7).

  When the pixel boundary Pi is a pixel boundary before the last pixel boundary included in the image frame (OP7: Yes), the processes OP1 to OP7 are repeated for the next pixel boundary.

  When the pixel boundary Pi is the last pixel boundary included in the image frame (OP7: No), the pixel boundary step accumulation unit 14 sends the accumulation result of the pixel boundary steps of each group to the determination value calculation unit 15. Output.

  When the cumulative result of the pixel boundary step in the image frame is input, the determination value calculation unit 15 subtracts the average of the cumulative value of the pixel boundary step of the non-block boundary group from the cumulative value of the pixel boundary step of the block boundary group. Then, a block noise detection determination value is calculated (OP8). The determination value calculation unit 15 outputs the block noise detection determination value to the block noise determination unit 16.

  The block noise determination unit 16 compares the block noise detection determination value with a plurality of block noise occurrence intensity thresholds to determine whether block noise has occurred in the processing target image frame and the block noise occurrence intensity ( OP9). The block noise determination unit 16 outputs the presence / absence of block noise and the block noise generation intensity in the processing target image frame.

  The block noise detection device 1A subtracts the average value of the pixel boundary steps of the non-block boundary group from the cumulative value of the pixel boundary steps of the block boundary group in the image frame to obtain a block noise detection determination value. As a result, the block noise detection determination value, which is the determination value of the block noise generation strength, becomes a value that approximates the cumulative value of steps due to block noise in the image frame, and the block noise generation strength can be detected with high accuracy.

  Further, the block noise detection device 1A obtains an estimated pixel value at the pixel boundary using higher-order prediction or linear prediction, and the difference between the absolute value of the pixel value of the two pixels forming the pixel boundary and the estimated pixel value The smaller of the absolute values is adopted as the pixel boundary step. By considering the estimated value of the pixel value at the pixel boundary, it is possible to accurately determine the level difference due to the pattern at each pixel boundary, and to determine the block noise detection judgment value calculated from the accumulated value of the pixel boundary level more accurately. Can do. As a result, the accuracy of detecting the block noise occurrence intensity is improved.

  Further, the block noise detection device 1A does not include a pixel boundary step having a sufficiently large threshold T1 or more due to a difference due to occurrence of block noise in the cumulative calculation of the pixel boundary step in the image frame. When the pixel boundary step is greater than or equal to the threshold T1, there is a high possibility that the step due to the pattern at the pixel boundary is large. By not including the pixel boundary step of the pixel boundary having a large step due to the pattern in the cumulative calculation of the pixel boundary step in the image frame, it is possible to reduce the influence of the image pattern on the block noise detection determination value. As a result, the accuracy of detecting the block noise occurrence intensity is improved.

  In the first embodiment, the block noise detection apparatus 1 </ b> A has been described with respect to an example in which block noise is detected for a moving image signal encoded and decoded using a pixel block. The block noise detection apparatus 1A can detect block noise not only for moving image signals but also for still image signals that are encoded and decoded using pixel blocks. For example, JPEG (Joint Photographic Experts Group) is used as an encoding compression method for encoding and decoding still images using pixel blocks.

  In the first embodiment, the block noise detection apparatus 1 </ b> A performs pixel processing in the horizontal direction, but may perform pixel processing in the vertical direction. In this case, the pixel boundary step calculation unit 13 buffers the pixels for one image frame, and starts processing after the pixels for one image frame are accumulated. The pixel boundary step calculation unit 13 obtains a pixel boundary step between the pixel boundaries of two pixels adjacent in the vertical direction. The pixel boundary step accumulation unit 14 includes a pixel boundary step at each pixel boundary in the image frame in each group of pixel boundary steps at each pixel boundary located on the horizontal line indicated by reference numerals d0 to d7 of the pixel block illustrated in FIG. And the pixel boundary step is accumulated in the classified group. The pixel boundary step accumulation unit 14 detects a block boundary group and a non-block boundary group from the accumulated value of the pixel boundary steps of the group, and outputs the block boundary group and the non-block boundary group 15 to the determination value calculation unit 15. Thereafter, similarly to the first embodiment, the determination value calculation unit 15 calculates the block noise detection determination value, and the block noise determination unit 16 detects the block noise occurrence intensity according to the block noise detection determination value.

  In the first embodiment, when obtaining the block noise detection determination value, the average of the accumulated values of the pixel boundary steps of each non-block boundary group is subtracted from the accumulated value of the pixel boundary steps of the block boundary group. Alternatively, the block noise detection determination value may be calculated by subtracting the accumulated value of the pixel boundary step of one non-block boundary group from the accumulated value of the pixel boundary step of the group boundary group.

Second Embodiment
In the second embodiment, the block noise detection apparatus extracts only a portion where the pixel value difference between the pixels at the same position in the processing target image frame and the immediately preceding image frame is large, performs cumulative calculation of the pixel boundary step, and performs block noise. A detection determination value is calculated. In the second embodiment, the description common to the first embodiment is omitted.

  FIG. 11 is a diagram illustrating a configuration example of a block noise detection system 100B in the second embodiment. In the second embodiment, the block noise detection system 100B includes a network 5, a storage device 6, a portable recording medium 7, an input unit 11, a decoding unit 12, and a block noise detection device 1B. The block noise detection apparatus 1B includes a pixel boundary step calculation unit 13, a pixel boundary step accumulation unit 14, a determination value calculation unit 15, a block noise determination unit 16, an interframe difference calculation unit 17, and a region detection unit 18.

  In the second embodiment, as in the first embodiment, the decoding unit 12 decodes a moving image signal to generate an image frame group, and sequentially includes pixels included in each image frame with a pixel boundary step calculation unit 13. Output to the inter-frame difference calculation unit 17.

The inter-frame difference calculation unit 17 has a buffer, and when a pixel is input from the decoding unit 12, the pixel is accumulated in the buffer. When the pixels for one image frame are accumulated in the buffer, the inter-frame difference calculation unit 17 divides the image frame into pixel blocks. The inter-frame difference calculation unit 17 buffers the image frame for at least a predetermined time, and holds the image frame to be processed and the image frame accumulated before that. The inter-frame difference calculation unit 17 associates the target image frame with the immediately preceding image frame, and calculates a difference absolute value of pixel values between the pixel in the target image frame and the pixel in the immediately preceding image frame that are located at the same position. To do. The inter-frame difference calculation unit 17 accumulates the absolute difference value of the pixel value between the pixel in the target image frame and the pixel in the immediately preceding image frame located at the same position in each pixel block. The accumulated value obtained by accumulating the absolute difference values of the pixel values of the pixels located at the same position in the target image frame and the immediately preceding image frame in each pixel block is hereinafter referred to as an inter-frame difference. The inter-frame difference calculation unit 17 outputs the calculated inter-frame difference of each pixel block between the calculated target image frame and the previous image frame to the region detection unit 18.

  When the size of the pixel block is unknown, the inter-frame difference calculation unit 17 does not operate until the pixel block size is detected by the pixel boundary step accumulation unit 14 as described in the first embodiment. The inter-frame difference calculation unit 17 starts operation after the pixel block size is detected by the pixel boundary step accumulation unit 14.

  The area detection unit 18 obtains, as an input, the inter-frame difference of each pixel block between the target image frame and the previous image frame. The area detection unit 18 extracts a pixel block having an inter-frame difference equal to or greater than an accumulation area threshold, and detects the pixel boundary step as an accumulation area.

  FIG. 12 is a diagram for explaining the accumulation target region detection processing. FIG. 12 shows an image frame with a pool with moving water in a stationary background.

  For example, when a moving image signal compressed and encoded by MPEG-2 is decoded, a stationary background portion such as a poolside floor in the example shown in FIG. Since the amount of data is small, block noise is less likely to occur. On the other hand, in a portion with a strong motion such as the water surface of the pool in the example shown in FIG. Since the inter-frame difference is large in the portion where the motion is intense, the portion where the inter-frame difference is large is likely to generate block noise.

  Therefore, the region detection unit 18 compares the inter-frame difference of each pixel block included in the target image frame with the cumulative target region threshold, and detects a pixel block whose inter-frame difference is equal to or greater than the cumulative target region threshold as the cumulative target region. . The accumulation target area detected in this way includes a plurality of pixel blocks. The area detection unit 18 outputs the detected accumulation target area of the pixel boundary step to the pixel boundary step calculation unit 13 and the block noise determination unit 16. The area detection unit 18 corresponds to a determination unit.

  The pixel boundary level calculating unit 13 buffers when a pixel is input from the decoding unit 12. When the pixel boundary step calculation unit 13 receives the accumulation target region of the pixel boundary step from the region detection unit 18 as an input, the pixel boundary step calculation unit 13 extracts the accumulation target region from the buffered target image frame, and extracts the pixel boundary step located in the accumulation target region. A pixel boundary step is calculated. The pixel boundary step calculation unit 13 outputs the calculated pixel boundary step at the pixel boundary in the accumulation target region to the pixel boundary step accumulation unit 14. Until the pixel block accumulating unit 14 detects the size of the pixel block, the pixel boundary step calculating unit 13 performs the same processing as in the first embodiment.

  The pixel boundary step accumulation unit 14 classifies the pixel boundary steps at the pixel boundary in the accumulation target region from the pixel boundary step calculation unit 13 into groups according to the positions in the pixel block of each pixel boundary, and the classified groups To accumulate the pixel boundary step. As a cumulative result, a cumulative value of the pixel boundary step of a group of the number of pixel block sizes is obtained. The pixel boundary step accumulation unit 14 outputs the accumulation result to the determination value calculation unit 15.

The determination value calculation unit 15 obtains, as an input, a cumulative value of pixel boundary steps of groups corresponding to the number of pixel block sizes as a cumulative result. The determination value calculation unit 15 calculates a block noise detection determination value by subtracting the average value of the pixel boundary step of the non-block boundary group from the cumulative value of the pixel boundary step of the block boundary group. The determination value calculation unit 15 outputs the block noise detection determination value to the block noise determination unit 16.

  The block noise determination unit 16 receives the block noise detection determination value from the determination value calculation unit 15 and the accumulation target region from the region detection unit 18 as inputs. The block noise determination unit 16 resets the values of the block noise occurrence intensity thresholds 1 to 3 according to the size of the accumulation target region. For example, the block noise determination unit 16 sets the block noise occurrence intensity thresholds 1 to 3 according to the ratio between the size of the image frame and the size of the accumulation target region. The block noise determination unit 16 compares the block noise detection determination value with the block noise generation intensity thresholds 1 to 3 to determine the block noise generation intensity.

  In the example shown in FIG. 12, when the block noise detection determination value is calculated based on the accumulated value of the pixel boundary step at the pixel boundary of the entire image frame, the water surface of the pool where the block noise is generated, and the block noise Pixel boundary steps are averaged with a stationary floor that has not occurred. Then, the pixel boundary step per unit area becomes small, and it becomes difficult to detect the accurate block noise generation intensity.

  On the other hand, the block noise detection apparatus 1B extracts only a region having a large interframe difference in which block noise is suspected (the water surface portion of the pool in FIG. 12) as a pixel boundary step accumulation target region. Further, the block noise detection device 1B accumulates only pixel boundary steps at the pixel boundary in the accumulation target region, calculates a block noise detection determination value based on the accumulated value, and blocks noise corresponding to the size of the accumulation target region. The block noise occurrence intensity is determined using the occurrence intensity threshold. This makes it possible to determine the block noise occurrence intensity with higher accuracy.

<Application example>
The block noise detection devices 1A and 1B of the first embodiment and the second embodiment are used as follows, for example.

  FIG. 13 is a diagram illustrating a configuration example of the block noise reduction system 200. The block noise reduction system 200 includes a network 5, a storage device 6, a portable recording medium 7, an input unit 11, a decoding unit 12, and a block noise reduction device 2. The block noise reduction device 2 controls the strength of filter processing for reducing block noise according to the block noise generation strength output from the block noise detection device 1A.

  The block reduction device 2 includes a block noise detection device 1A and a block noise reduction unit 21. The block noise detection device 1 </ b> A outputs the image frame and the block noise occurrence intensity to the block noise reduction unit 21.

  The block noise reduction unit 21 receives an image frame and a block noise generation intensity as inputs. The block noise reduction unit 21 includes a block noise reduction filter 1, a block noise reduction filter 2, and a block noise reduction filter 3, and reduces block noise of the image frame according to the block generation intensity. The block noise reduction filters 1 to 3 are, for example, low-pass filters. The smoothing strength of the block noise reduction filter is in the order of block noise reduction filter 3> block noise reduction filter 2> block noise reduction filter 1.

The block noise reduction unit 21 outputs the target image frame as it is when the detection result of the block noise occurrence intensity that no block noise is generated is input. When the block noise generation intensity is weak, the block noise reduction unit 21 uses the block noise reduction filter 1 to perform block noise reduction processing of the target image frame. When the block noise generation intensity is medium, the block noise reduction unit 21 uses the block noise reduction filter 2 to perform block noise reduction processing of the target image frame. When the block noise generation intensity is strong, the block noise reduction unit 21 performs block noise reduction processing of the target image frame using the block noise reduction filter 3.

  In this application example, the block noise detection device 1A is used as the block noise reduction device 2. However, the block noise reduction device 2 uses the block noise detection device 1B in the same manner as the block noise detection device 1A. Can do.

<Hardware configuration of block noise detection device>
Information processing devices (computers) can be applied to the block noise detection devices 1A and 1B. The information processing apparatus can be realized using, for example, a general-purpose computer such as a personal computer or a dedicated computer that performs block noise detection.

  The information processing apparatus includes an interface device with peripheral devices such as a processor, a main storage device, an input device, an output device, a secondary storage device, and a communication interface device. The main storage device and the secondary storage device are computer-readable recording media.

  In the information processing device, the processor loads the program stored in the recording medium into the work area of the main storage device and executes it, and the peripheral device is controlled through the execution of the program, thereby realizing a function that meets a predetermined purpose. can do.

  The processor is, for example, a CPU (Central Processing Unit) or a DSP (Digital Signal Processor). The main storage device includes, for example, a ROM (Read Only Memory) and a RAM (Random Access Memory).

The secondary storage device is, for example, an EPROM (Erasable Programmable).
ROM), or hard disk drive (Hard Disk Drive).
Further, the secondary storage device can include a removable medium, that is, a portable recording medium. The removable media is, for example, a USB (Universal Serial Bus) memory, or a disc recording medium such as a CD (Compact Disc) or a DVD (Digital Versatile Disc).

  The communication interface device is connected to a wired network and a wireless network. The communication interface device is, for example, a LAN (Local Area Network) interface board or a wireless communication circuit for wireless communication.

  Further, the peripheral device includes an input device such as a keyboard and a pointing device, and an output device such as a display device and a printer. The input device may include an audio input device such as a microphone. The output device may include an audio output device such as a speaker.

  The computer used as the block noise detection device controls the peripheral device through execution of the block noise detection program on the recording medium by the processor, whereby the input unit 11, the decoding unit 12, the pixel boundary step calculation unit 13, the pixel Functions as the boundary step accumulation unit 14, the determination value calculation unit 15, the block noise determination unit 16, the interframe difference calculation unit 17, and the region detection unit 18 are realized.

  The block noise detection devices 1A and 1B include an input unit 11, a decoding unit 12, a pixel boundary step calculation unit 13, a pixel boundary step accumulation unit 14, a determination value calculation unit 15, a block noise determination unit 16, and an interframe difference calculation unit. 17. It can also be realized as an electronic circuit and an integrated circuit having the function of the area detection unit 18.

1A, 1B Block noise detection device 2 Block noise reduction device 11 Input unit 12 Decoding unit 13 Pixel boundary step calculation unit 14 Pixel boundary step accumulation unit 15 Determination value calculation unit 16 Block noise determination unit 17 Interframe difference calculation unit 18 Region detection unit 21 Block Noise Reduction Unit 100A, 100B Block Noise Detection System 200 Block Noise Reduction System

Claims (6)

A target image frame obtained by receiving image data encoded in units of pixel blocks each including a plurality of pixels in which at least one input image frame is arranged in a grid, and decoding each of the image data in units of pixel blocks In the block noise detection device for detecting block noise in
A level difference calculating unit that calculates a difference in pixel value between adjacent pixels in the target image frame as a pixel boundary level of a pixel boundary between the pixels;
The calculated pixel boundary step is classified into one of a plurality of groups corresponding to the position of the pixel boundary in each pixel block, and the pixel boundary step is accumulated for each classified group. And
Among the plurality of groups, a block boundary group corresponding to a position where the pixel boundary is a boundary between adjacent pixel blocks is specified, and a value other than the block boundary group is determined from the accumulated value of the pixel boundary step of the block boundary group. A value obtained by subtracting an accumulated value of pixel boundary steps of one group of groups or an average value of accumulated values of pixel boundary steps of a plurality of groups other than the block boundary group is calculated as a block noise determination value. A calculation unit;
A detection unit for detecting a block noise occurrence state in the target image frame according to the block noise determination value;
A block noise detection apparatus including: The level difference calculating unit is configured to calculate a difference value between a pixel value of one of two adjacent pixels forming a pixel boundary and a pixel value of the other pixel, and one direction of the pixel boundary including the one pixel. The pixel boundary estimated from the relationship between the pixel value at the pixel boundary estimated from the relationship between the pixel values of the plurality of pixels on the side and the pixel value of the plurality of pixels on the other direction side of the pixel boundary including the other pixel The smaller one of the difference values from the pixel value is calculated as the pixel boundary step.
The block noise detection apparatus according to claim 1. The accumulation unit accumulates the pixel boundary step in a classified group when the pixel boundary step is smaller than a predetermined threshold value, and does not accumulate when the pixel boundary step is a predetermined threshold value or more. The block noise detection device according to 1 or 2. A determination unit that determines a pixel boundary that is an accumulation target of the pixel boundary step based on a difference in pixel values between pixels at the same position in the target image frame and an image frame immediately before the target image frame; Prepared,
4. The block noise detection device according to claim 1, wherein the accumulation unit accumulates pixel boundary steps of the pixel boundary to be accumulated. 5. A target image frame obtained by receiving image data encoded in units of pixel blocks each including a plurality of pixels in which at least one input image frame is arranged in a grid, and decoding each of the image data in units of pixel blocks A block noise detection device for detecting block noise in
Calculating a difference in pixel value between adjacent pixels in the target image frame as a pixel boundary step of a pixel boundary between the pixels,
Classifying each calculated pixel boundary step into one of a plurality of groups corresponding to the position of the pixel boundary in each pixel block, and accumulating each pixel boundary step for each classified group When,
Among the plurality of groups, a block boundary group corresponding to a position where a pixel boundary is a boundary between adjacent pixel blocks is specified, and a group other than the block boundary group is determined from a cumulative value of pixel boundary steps of the block boundary group A value obtained by subtracting an accumulated value of pixel boundary steps of one of the groups or an average value of accumulated values of pixel boundary steps of a plurality of groups other than the block boundary group as a block noise determination value When,
A detection step of detecting a block noise occurrence state in the target image frame according to the block noise determination value;
Block noise detection method to execute. A target image frame obtained by receiving image data encoded in units of pixel blocks each including a plurality of pixels in which at least one input image frame is arranged in a grid, and decoding each of the image data in units of pixel blocks In block noise detection device that detects block noise in
Calculating a difference in pixel value between adjacent pixels in the target image frame as a pixel boundary step of a pixel boundary between the pixels,
Classifying each calculated pixel boundary step into one of a plurality of groups corresponding to the position of the pixel boundary in each pixel block, and accumulating each pixel boundary step for each classified group When,
Among the plurality of groups, a block boundary group corresponding to a position where a pixel boundary is a boundary between adjacent pixel blocks is specified, and a group other than the block boundary group is determined from a cumulative value of pixel boundary steps of the block boundary group A value obtained by subtracting an accumulated value of pixel boundary steps of one of the groups or an average value of accumulated values of pixel boundary steps of a plurality of groups other than the block boundary group as a block noise determination value When,
Detecting a block noise occurrence state in the target image frame according to the block noise determination value;
Block noise detection program to execute.

Images