JP4475196B2 - Block noise detection device and block noise detection program - Google Patents

Description

  The present invention relates to a block noise detection apparatus and a block noise detection program, and in particular, detects block noise generated at a block boundary when encoding and decoding when a video signal is compressed and transmitted and recorded as a digital image. The present invention relates to a block noise detection apparatus and a block noise detection program for causing a computer to detect block noise.

  As a compression encoding method when digitally compressing a video signal, there is a method in which a plurality of pixels adjacent in a horizontal direction and a vertical direction of a video signal to be encoded are set as one rectangular block and compression encoding is performed in units of this block. Are known. The video signal that has been compression-encoded by such a compression-encoding method is subjected to a decoding process in units of rectangular blocks after passing through a recording medium or a transmission path, so that the original video signal is restored.

  In such a compression encoding and decoding method, if the compression rate is increased when it is desired to reduce the information amount of the video signal on the recording medium or the transmission path, a gradation difference occurs between the rectangular blocks. In general, noise due to this gradation difference is called block noise. As a method of detecting block noise generated at such block boundaries, there is generally a method of detecting block boundaries based on information of the video compression data decoding apparatus.

  Also, a block noise detection method when there is no block boundary information of a video compression data decoding apparatus is known in the art (see, for example, Patent Documents 1 and 2). In the conventional block noise detecting device described in Patent Document 1, the high-frequency components in the horizontal direction and the vertical direction of the video signal to be detected are extracted by the horizontal high-pass filter and the vertical high-pass filter, respectively. After the absolute value is converted into an absolute value, the horizontal cumulative adder and vertical cumulative adder cumulatively add the respective absolute value signals to have a horizontal one-dimensional signal having a peak value in the horizontal period and a peak value in the vertical period. A vertical one-dimensional signal is obtained.

  Then, a horizontal peak position is detected at horizontal block size intervals by the horizontal peak detection unit based on the horizontal one-dimensional signal, and a vertical peak detection unit is detected by the vertical peak detection unit based on the vertical one-dimensional signal. After detecting vertical peak positions at block size intervals, block boundary portions are detected based on these peak position detection signals, and are detected as locations where block noise occurs.

  Further, in the conventional block noise detection device described in Patent Document 2, a block boundary difference value that is an absolute value of a difference between two pixels adjacent to each other in the horizontal direction or the vertical direction across the target point, and between adjacent pixels The block boundary where block noise occurs is detected by using the adjacent difference value which is the difference value of.

JP 2000-50275 A JP 2002-232889 A

  However, in the above-described method for detecting the block boundary based on the information of the video compression data decoding apparatus, block noise cannot be detected without the block boundary information from the video compression data decoding apparatus. For example, when a video signal having block noise is input from an external input in a television receiver, block noise generated at the block boundary cannot be detected because there is no block boundary information of the video compression data decoding apparatus.

  On the other hand, in the conventional block noise detection device described in Patent Document 1, block noise can be detected without the block boundary information from the video compression data decoding device, but a cumulative adder is used for block boundary periodicity detection. Therefore, when the input video signal amount increases, the circuit scale of the cumulative adder increases.

  Further, in the conventional block noise detection device described in Patent Document 2, block noise can be detected without the block boundary information from the video compression data decoding device, but the block boundary difference is calculated from the average value of adjacent difference values. Since the value is calculated, there is a problem that an original video signal or random noise is recognized as a block boundary difference value.

  The present invention has been made in view of the above points. Even if there is no block boundary information from the video compression data decoding apparatus, the block noise can be accurately detected from the input video signal in which block noise is generated. An object of the present invention is to provide a noise detection device and a block noise detection program.

In order to achieve the above object, according to a first aspect of the present invention, a coded video signal obtained by compressing and coding a video signal in units of blocks composed of a plurality of pixels is converted into a decoded video signal when decoded in units of blocks. When block noise occurs, in a block noise detection device that detects a block boundary where the block noise has occurred, for each of a plurality of pixels in the horizontal period of the decoded video signal, a difference absolute value with an adjacent pixel is calculated. Are arranged in time series within a predetermined period in the horizontal direction determined by the number of horizontal pixels m (m is a natural number of 2 or more) in the block unit, and extraction means for extracting the adjacent pixel difference absolute value at the pixel position corresponding to each of the pixels Adjacent pixel difference absolute values at m pixel positions are ranked in descending order, and one preset image within a predetermined period in the horizontal direction is set. When the order of the adjacent pixel difference absolute values at the position target pixel position which is one or more predefined predetermined rank that includes at least up to the rank, the adjacent pixel difference absolute value at the target pixel position in the horizontal direction by a predetermined time period When the rank is not the predetermined rank, the process of outputting the numerical value 0 as the representative value is executed while shifting the horizontal predetermined period by one pixel in the horizontal direction, and the representative value in each horizontal predetermined period As a representative value for each pixel position in the horizontal period, and a representative value output for each pixel position in the horizontal period as an input from the ranking means. When the m pixel positions arranged in a predetermined period in the horizontal direction are the first to mth pixel positions in time series in order from the front, An average value calculating means for calculating an average value of representative values at pixel positions in the same order within a predetermined period in the horizontal direction, and an average value at each of the first to mth pixel positions calculated by the average value calculating means Among these, when the largest average value is equal to or larger than a predetermined value obtained based on the second largest average value, the pixel position corresponding to the largest average value is extracted, and the pixel position corresponding to the extracted pixel position within the horizontal period is extracted. A fixed value is output from the ranking unit at a detection unit that detects all pixel positions appearing in a cycle unit of a predetermined period in the horizontal direction as a pixel position having a block boundary, and a pixel position at which the block boundary is detected by the detection unit. Block boundary to detect the pixel position as a horizontal block boundary where block noise occurs when a fixed value is not output And detecting means.

In the present invention, adjacent pixel difference absolute values at m pixel positions arranged in time series within a predetermined horizontal direction determined by the number of horizontal pixels m of the block are ranked in descending order, and the predetermined horizontal direction is determined. When the rank of the adjacent pixel difference absolute value at the target pixel position which is one pixel position set in advance during the period is at least one predetermined predetermined order including at least the maximum rank, the adjacent position at the target pixel position The process of outputting the pixel difference absolute value as a representative value for a predetermined period in the horizontal direction and outputting the numerical value 0 as a representative value when the rank is not a predetermined rank is executed while shifting the predetermined period in the horizontal direction by one pixel at a time. The average value of the m pixel positions in the same order among the first to m-th pixel positions within a predetermined horizontal direction obtained by the ranking is calculated. A horizontal direction corresponding to the pixel position corresponding to the largest average value when a value is calculated and the largest average value among these average values is equal to or greater than a predetermined value obtained based on the second largest average value When all the pixel positions appearing in the period unit of the predetermined period in the horizontal direction within the predetermined period are set as pixel positions having a block boundary, and the adjacent pixel difference absolute value is output as a representative value from the ranking means at the pixel position. Since the pixel position is detected as the horizontal block boundary where the block noise is generated, the horizontal block boundary where the block noise is generated can be detected using only the input decoded video signal.

  In order to achieve the above object, in the second invention, the representative value output from the ranking means is a value between a first threshold value and a second threshold value that are different from each other. When the representative value output from the ranking means satisfies the condition, the output value is output to the average value calculation means and the horizontal boundary detection means, and the condition is not satisfied Is further characterized by further comprising determination means for cutting off output to the average value calculating means and block boundary detecting means of the representative values output from the ranking means.

  In the present invention, when the representative value output from the ranking means is not a value within the range between the first threshold value and the second threshold value, the average value calculating means and block boundary detection of the representative values output from the ranking means Since the output to the means is cut off, when the representative value output from the ranking means is not a value within the range between the first threshold value and the second threshold value, the above representative value is much higher than the block noise. Therefore, it can be determined that the pixel position corresponding to the representative value is not a horizontal block boundary position where block noise occurs.

  In order to achieve the above object, the third aspect of the invention provides a decoded video signal obtained by decoding an encoded video signal obtained by compressing and encoding a video signal in units of blocks composed of a plurality of pixels in units of blocks. In the block noise detecting apparatus for detecting the block boundary where the block noise occurs when the block noise occurs in the signal, the horizontal direction of the first invention is changed to the vertical direction. In the third aspect of the invention, the block boundary in the vertical direction where block noise is generated can be detected using only the input decoded video signal.

  In order to achieve the above object, according to a fourth aspect, in the third aspect, the representative value output from the ranking means is between a first threshold value and a second threshold value that are different from each other. If the representative value output from the ranking means satisfies the condition, the output value is output to the average value calculation means and the vertical boundary detection means. When the condition is not satisfied, the average value calculating means for the representative value output from the ranking means and the determination means for cutting off the output to the block boundary detecting means are further provided.

  In the present invention, when the representative value output from the ranking means is not a value within the range between the first threshold value and the second threshold value, the average value calculating means and block boundary detection of the representative values output from the ranking means Since the output to the means is cut off, when the representative value output from the ranking means is not a value within the range between the first threshold value and the second threshold value, the above representative value is much higher than the block noise. Therefore, it can be determined that the pixel position corresponding to the representative value is not the block boundary position in the vertical direction where the block noise is generated.

  In order to achieve the above object, a block noise detection program according to a fifth aspect of the present invention decodes an encoded video signal obtained by compressing and encoding a video signal in units of blocks composed of a plurality of pixels in units of blocks. A block noise detection program for causing a computer to detect a horizontal block boundary in which block noise has occurred when block noise occurs in the decoded video signal at the time of block decoding by the block noise detection apparatus of the first invention Noise detection is performed by a computer.

  In order to achieve the above object, a block noise detection program according to a sixth aspect of the invention decodes an encoded video signal obtained by compressing and encoding a video signal in units of blocks composed of a plurality of pixels in units of blocks. When block noise occurs in the decoded video signal at the time, a block noise detection program for causing a computer to detect a block boundary in the vertical direction in which the block noise has occurred, and the block noise detection apparatus according to the third invention Noise detection is performed by a computer.

  According to the present invention, the horizontal block boundary or the vertical block boundary where block noise is generated can be detected using only the input decoded video signal. Even when there is no block noise, it is possible to detect block noise generated at the block boundary.

  Further, according to the present invention, when the representative value output from the ranking unit is not a value within the range between the first threshold value and the second threshold value, block noise occurs at the pixel position corresponding to the representative value. Since it is determined that the block boundary position is not in the horizontal or vertical direction, the original decoded video whose representative value output from the ranking means is not a value within the range between the first threshold value and the second threshold value. It is possible to prevent erroneous detection of a signal as a horizontal boundary or a vertical boundary of a block in which block noise occurs.

  Next, the best mode for carrying out the present invention will be described with reference to the drawings. FIG. 1 shows a block diagram of a first embodiment of a block noise detection apparatus according to the present invention. The present embodiment is a device that detects block noise in the horizontal direction, and includes a vertical low-pass filter (LPF) 11, an adjacent pixel difference absolute value extraction circuit 12 within a predetermined horizontal direction, and a pixel of interest by ranking adjacent pixel difference values. The circuit includes a difference absolute value extraction circuit 13, an average value calculation circuit 14 based on a block horizontal pixel period, a block horizontal pixel periodicity detection circuit 15, and a block horizontal boundary detection circuit 16.

  The vertical LPF 11 is a decoded video signal in which a video signal compressed and encoded in units of blocks by a certain compression encoding method is supplied to a decoding circuit (not shown) via a predetermined transmission path and decoded in units of blocks. Is extracted, and the low frequency component in the vertical direction of the decoded video signal is extracted. This is to make it easier to detect the block horizontal boundary, so that all the horizontal components remain and the high frequency components in the vertical direction are attenuated.

  The vertical LPF 11 is configured as shown in the block diagram of FIG. 3, for example, and sequentially delays an input decoded video signal line by line by four cascaded line memories 21, 22, 23 and 24. The 1-line delayed decoded video signal, the 2-line delayed decoded video signal, and the 3-line delayed decoded video signal output from the line memories 21, 22, and 23 are respectively multiplied by predetermined multiplier coefficients by multipliers 25, 26, and 27, respectively. After that, the signal is supplied to the adder 28, where the non-delay decoded video signal and the 4-line delayed decoded video signal output from the line memory 24 are added and synthesized.

  The synthesized video signal obtained by adding and synthesizing by the adder 28 is multiplied by a predetermined coefficient by the multiplier 29 and level-adjusted and then output. The signal output from the adder 29 is a low-frequency component in the vertical direction obtained by calculating the vertical pixel values of five adjacent lines of the input decoded video signal. It is supplied to the adjacent pixel difference absolute value extraction circuit 12 within a predetermined period.

  The adjacent pixel difference absolute value extraction circuit 12 within a predetermined period in the horizontal direction extracts an adjacent pixel difference absolute value within a predetermined period in the horizontal direction of the input signal (a low frequency component in the vertical direction of the input video signal). That is, the adjacent pixel difference absolute value extraction circuit 12 sets the horizontal horizontal predetermined period to 8 pixels when the number of block horizontal pixels of the block noise is 8 pixels, for example, and as shown in FIG. The absolute value of the difference value between each of the eight pixels P1 to P8 arranged in time series in the horizontal direction and the pixels P0 to P7 immediately adjacent to the left is calculated. As shown in FIG. The absolute value of the difference value (adjacent pixel difference absolute value) is calculated.

  In FIG. 5, 1 is the absolute difference value between the pixels P1 and P0 in FIG. 4, 2 is the absolute difference value between the pixels P2 and P1 in FIG. 4, and 3 is the absolute difference between the pixels P3 and P2 in FIG. 4 is an absolute difference value between the pixels P4 and P3 in FIG. 4, 5 is an absolute difference value between the target pixel P5 and the pixel P4 in FIG. 4, and 6 is an absolute difference value between the pixel P6 and the target pixel P5 in FIG. , 7 are absolute differences between the pixels P7 and P6 in FIG. 4, and 8 are absolute differences between the pixels P8 and P7 in FIG.

  The adjacent pixel difference absolute value extraction circuit 12 shown in FIG. 1 performs a process of calculating the absolute value of the difference value between the eight pixels in the predetermined period in the horizontal direction and the pixel immediately adjacent to the left in the horizontal direction. Execute while shifting pixel by pixel. The eight adjacent pixel difference absolute values within the predetermined period in the horizontal direction calculated by the adjacent pixel difference absolute value extraction circuit 12 in this way are the target pixel difference absolute value extraction circuit 13 based on the ranking of the adjacent pixel difference values. To be supplied.

  The target pixel difference absolute value extraction circuit 13 ranks the eight adjacent pixel difference absolute values indicated by 1 to 8 in FIG. 5 in descending order, and is approximately in the middle position among the eight adjacent pixel difference absolute values. When the difference absolute value of the target pixel (target pixel difference absolute value) indicated by 5 in FIG. 5 is a predetermined order, for example, when the order is first or second, the target pixel difference absolute value is represented. Output as a value. Further, the pixel-of-interest difference absolute value extraction circuit 13 sets a predetermined fixed value “0” when the pixel-of-interest absolute value is not the above-mentioned predetermined order, for example, when it is not the first or second value from the largest. Output as a representative value.

  The processing by the pixel-of-interest absolute value extraction circuit 13 is performed so that eight adjacent pixel difference absolute values input to the pixel-of-interest difference absolute value extraction circuit 13 Since it is eight adjacent pixel difference absolute values obtained by executing the process of calculating the absolute value of the difference value between the eight pixels and the pixel immediately adjacent to the pixel by shifting one pixel at a time in the horizontal direction, This is a process of outputting the above representative value for eight adjacent pixel difference absolute values inputted while shifting one pixel at a time in the horizontal direction.

  The average value calculation circuit 14 based on the block horizontal pixel period in FIG. 1 receives the representative value from the above target pixel difference absolute value extraction circuit 13, that is, the target pixel difference absolute value or “0” as an input, Based on the block horizontal pixel period, the average value is calculated. For example, when the block horizontal pixel period is 8 pixels, the average value calculation circuit 14 based on the block horizontal pixel period includes eight average value calculators in parallel, and is output from the above noted pixel difference absolute value extraction circuit 13. A value (absolute value of pixel difference of interest or “0”) is cyclically supplied to the eight average value calculators in the order of the pixels in the horizontal direction, and in a unit of period in the horizontal direction for all pixels. The average value of the target pixel input values is calculated.

  That is, as shown in FIG. 6, the average value calculation circuit 14 based on the block horizontal pixel period looks at all pixels in one frame for each pixel from the upper left of the screen of one frame, and looks at the target pixel input value for each pixel. The 8 pixel average absolute values or fixed values “0” extracted at the pixel positions of the 8 pixels indicated by 1 to 8 are individually averaged (average value based on the block horizontal pixel period) by 8 average value calculators. ) Is calculated. That is, among the eight pixels in the horizontal pixel period indicated by 1 to 8 in FIG. 6, an average value is calculated between the values of pixels having the same number (same order), and as a result, a total of eight average values are obtained. (Average value based on the block horizontal pixel period) is calculated. The average value is not calculated for each frame, but the average value is calculated for every eight input pixel units in a predetermined period in the horizontal direction as described above.

  The block horizontal pixel periodicity detection circuit 15 in FIG. 1 detects the periodicity of the block noise boundary from the average value of the eight block horizontal pixel periods calculated by the average value calculation circuit 14 based on the block horizontal pixel period. That is, the block horizontal pixel periodicity detection circuit 15 ranks the average values of the eight input block horizontal pixel periods, and when the first largest average value is n times or more than the second largest average value (n Is an arbitrary number), and detects that the portion of the average value that is the first largest in 8 horizontal pixel periods (the pixel position where the first largest average value is obtained) is the block noise boundary period. When the first largest average value is not more than n times the second largest average value, it is detected that none of the eight pixels indicated by 1 to 8 in FIG. 6 is a block noise boundary period.

  The block horizontal boundary detection circuit 16 in FIG. 1 is a horizontal pixel periodicity detection signal indicating that one of the eight pixels 1 to 8 shown in FIG. 6 from the block horizontal pixel periodicity detection circuit 15 is a block noise boundary period. When the pixel of interest corresponds to the block noise boundary period and the value output from the pixel-of-interest difference absolute value extraction circuit 13 is a value other than 0, the pixel position is the occurrence of block noise. It is detected that this is a horizontal block boundary (that is, a block horizontal boundary).

  In this way, according to the present embodiment, it is possible to detect a horizontal block boundary where block noise has occurred without requiring block noise boundary information of the video compression data decoding apparatus. Further, according to the present embodiment, since the horizontal pixel periodicity is detected from the average value without using the cumulative adder for detecting the block boundary periodicity, the cumulative adder is used even if the input video signal amount increases. In comparison, the circuit scale is small.

  Next, a second embodiment of the present invention will be described. FIG. 2 shows a block diagram of a second embodiment of a block noise detection apparatus according to the present invention. In the figure, the same components as those in FIG. The second embodiment shown in FIG. 2 is characterized in that an attention pixel difference absolute value determination circuit 17 based on two threshold values is added to the first embodiment shown in FIG.

  In FIG. 2, the pixel-of-interest difference absolute value determination circuit 17 using two thresholds provides two thresholds A and B (where threshold A> threshold B, A and B are arbitrary numbers), and pixel-of-interest difference absolute value extraction When the target pixel difference absolute value supplied from the circuit 13 is satisfied and an inequality of threshold A ≧ target pixel difference absolute value ≧ threshold B is satisfied, the target pixel difference absolute value supplied from the target pixel difference absolute value extraction circuit 13 is satisfied. The value is output as it is, and when the above inequality is not satisfied (when the target pixel difference absolute value is larger than the threshold A or smaller than the threshold B), “0” is output.

  The signals output from the target pixel difference absolute value determination circuit 17 based on the two threshold values are supplied to the average value calculation circuit 14 and the block horizontal boundary detection circuit 16 based on the block horizontal pixel period, respectively. Thus, when the target pixel difference absolute value is larger than the threshold A or smaller than the threshold B, it is determined that the signal is other than block noise, and the target pixel difference absolute value output from the target pixel difference absolute value extraction circuit 13 is Since it is not sent to the subsequent stage, it is possible to prevent false detection that the original video signal is detected as a horizontal block boundary (block horizontal boundary) where block noise has occurred. It can be detected.

  Next, a third embodiment of the present invention will be described. FIG. 7 shows a block diagram of a third embodiment of a block noise detecting apparatus according to the present invention. The present embodiment is a device that detects block noise in the vertical direction, and includes a horizontal low-pass filter (LPF) 31, an adjacent pixel difference absolute value extraction circuit 32 within a predetermined period in the vertical direction, and a pixel of interest by ranking adjacent pixel difference values. The circuit includes a difference absolute value extraction circuit 33, an average value calculation circuit 34 based on a block vertical pixel period, a block vertical pixel periodicity detection circuit 35, and a block vertical boundary detection circuit 36.

  The horizontal LPF 31 is a decoded video signal in which a video signal compressed and encoded in units of blocks by a certain compression encoding method is supplied to a decoding circuit (not shown) via a predetermined transmission path and decoded in units of blocks. Is extracted, and the low frequency component in the horizontal direction of the decoded video signal is extracted. This is to make it easier to detect the block vertical boundary, so that all the components in the vertical direction are left and the high frequency component in the horizontal direction is attenuated.

  The horizontal LPF 31 is configured as shown in the block diagram of FIG. 9, for example, and sequentially delays the input decoded video signal by one pixel period by four cascaded one-pixel delay units 41, 42, 43, and 44. The one-pixel period delayed decoded video signal, the two-pixel period delayed decoded video signal, and the three-pixel period delayed decoded video signal output from the one-pixel delay units 41, 42, 43, and 44 are respectively separated by multipliers 45, 46, and 47, respectively. After being multiplied by a predetermined multiplication coefficient, it is supplied to the adder 48, where the non-delay decoded video signal and the 4-pixel period delayed decoded video signal output from the 1-pixel delay unit 44 are added and combined.

  The synthesized video signal obtained by adding and synthesizing by the adder 48 is multiplied by a predetermined coefficient by a multiplier 49 and level-adjusted and output. The signal output from the adder 49 is a horizontal low-frequency component obtained by calculating the horizontal pixel values of the adjacent five pixel periods of the input decoded video signal. The direction is supplied to the adjacent pixel difference absolute value extraction circuit 32 within a predetermined period.

  The adjacent pixel difference absolute value extraction circuit 32 within a predetermined period in the vertical direction extracts an adjacent pixel difference absolute value within a predetermined period in the vertical direction of the input signal (a low frequency component in the horizontal direction of the input video signal). That is, the adjacent pixel difference absolute value extraction circuit 32 sets the vertical vertical period to 8 pixels when the number of vertical pixels of a block having a block boundary where block noise occurs is 8 pixels, for example, as shown in FIG. Then, the absolute value of the difference value between each of the eight pixels D1 to D8 arranged in time series in the vertical direction in this vertical direction and the pixels D0 to D7 immediately adjacent thereto is calculated. The absolute value of the pixel difference value (adjacent pixel difference absolute value) is calculated.

  The adjacent pixel difference absolute value extraction circuit 32 shown in FIG. 7 performs a process of calculating the absolute value of the difference value between the eight pixels in the predetermined period in the vertical direction and the pixel immediately above it in the vertical direction. Execute while shifting pixel by pixel. In this way, the eight adjacent pixel difference absolute values calculated in the adjacent pixel difference absolute value extraction circuit 32 within the predetermined period in the vertical direction are the target pixel difference absolute value extraction circuit based on the ranking of the adjacent pixel difference values. 33.

  The target pixel difference absolute value extraction circuit 33 ranks the eight adjacent pixel difference absolute values inputted in descending order, and the target pixel difference indicated by D5 in FIG. When the pixel difference absolute value (target pixel difference absolute value) has a predetermined order, for example, when the order is first or second, the target pixel difference absolute value is output as a representative value. Further, the pixel-of-interest difference absolute value extraction circuit 33 sets a predetermined fixed value “0” when the pixel-of-interest absolute value is not the above-mentioned predetermined order, for example, when it is not the first or second value from the largest. Output as a representative value.

  The processing by the pixel-of-interest absolute value extraction circuit 33 is performed by the eight adjacent pixel-difference absolute values input to the pixel-of-interest-difference absolute value extraction circuit 33 in the vertical-direction predetermined period in the adjacent-pixel difference absolute value extraction circuit 32. Since it is the eight adjacent pixel difference absolute values obtained by executing the process of calculating the absolute value of the difference value between the eight pixels and the pixel immediately above it by shifting one pixel at a time in the vertical direction, This is a process of outputting the above representative value for eight adjacent pixel difference absolute values inputted while shifting one pixel at a time in the vertical direction.

  The average value calculation circuit 34 based on the block vertical pixel period in FIG. 7 receives the target pixel difference absolute value or “0” from the target pixel difference absolute value extraction circuit 33 as an input, and based on the input value, the block vertical pixel period The average value is calculated by. For example, when the block vertical pixel cycle is 8 pixels, the average value calculation circuit 34 based on the block vertical pixel cycle uses the pixel-of-interest absolute value or “0” from the pixel-of-interest absolute value extraction circuit 33 as vertical 8 pixels. The average value is calculated with the period.

  That is, as shown in FIG. 11, the average value calculation circuit 34 based on the block vertical pixel period looks at the pixel-of-interest input value for each pixel in the vertical direction from the upper left of the screen of one frame for every pixel. 11, the average value (average value based on the block vertical pixel period) of the pixel-of-interest difference absolute values extracted at the pixel positions of the eight pixels indicated by 1 to 8 in FIG. 11 is calculated separately by the eight average value calculators. . That is, among the eight pixels in the vertical pixel period indicated by 1 to 8 in FIG. 11, the average value is calculated between the values of the pixels having the same number (same order), and as a result, a total of eight average values are obtained. (Average value based on the block vertical pixel period) is calculated. The average value is not calculated for each frame, but is divided for every eight input pixel units in a predetermined period in the vertical direction as described above.

  The block vertical pixel periodicity detection circuit 35 in FIG. 7 detects the periodicity of the block noise boundary from the average value based on the eight block vertical pixel periods calculated by the average value calculation circuit 34 based on the block vertical pixel period. That is, the block vertical pixel periodicity detection circuit 35 ranks the average values of the eight input block vertical pixel periods, and when the first largest average value is n times or more than the second largest average value (n Is an arbitrary number), and it is detected that the portion of the average value that is the largest in eight periods (the pixel position corresponding to the first largest average value) is the block noise boundary period. When the first largest average value is not n times or more than the second largest average value, it is detected that none of the eight pixels indicated by 1 to 8 in FIG. 11 is a block noise boundary period.

  The block vertical boundary detection circuit 36 in FIG. 7 is a vertical pixel periodicity detection signal indicating that one of the eight pixels 1 to 8 shown in FIG. 11 from the block vertical pixel periodicity detection circuit 35 is a block noise boundary period. When the pixel of interest corresponds to the block noise boundary period and the value output from the pixel-of-interest difference absolute value extraction circuit 33 is a value other than 0, the pixel position is the occurrence of block noise. Detects that the block is a vertical boundary.

  In this way, according to the present embodiment, it is possible to detect a block boundary in the vertical direction in which block noise occurs without requiring block noise boundary information of the video compression data decoding apparatus. Further, according to the present embodiment, since the vertical pixel periodicity is detected from the average value without using the cumulative adder for detecting the block boundary periodicity, the cumulative adder is used even if the input video signal amount increases. In comparison, the circuit scale is small.

  Next, a fourth embodiment of the present invention will be described. FIG. 8 shows a block diagram of a fourth embodiment of a block noise detecting apparatus according to the present invention. In the figure, the same components as those in FIG. 7 are denoted by the same reference numerals, and the description thereof is omitted. The fourth embodiment shown in FIG. 8 is characterized in that an attention pixel difference absolute value determination circuit 37 based on two thresholds is added to the third embodiment shown in FIG.

  In FIG. 8, the pixel-of-interest difference absolute value determination circuit 37 based on two thresholds provides two thresholds A and B (where threshold A> threshold B, A and B are arbitrary numbers), and pixel-of-interest difference absolute value extraction. When the target pixel difference absolute value supplied from the circuit 33 is satisfied and the inequality of threshold A ≧ target pixel difference absolute value ≧ threshold B is satisfied, the target pixel difference absolute value supplied from the target pixel difference absolute value extraction circuit 33 is satisfied. The value is output as it is, and when the above inequality is not satisfied (when the target pixel difference absolute value is larger than the threshold A or smaller than the threshold B), “0” is output.

  Signals output from the target pixel difference absolute value determination circuit 37 based on the two threshold values are supplied to the average value calculation circuit 34 and the block vertical boundary detection circuit 36 based on the block vertical pixel period, respectively. Thereby, when the value output from the target pixel difference absolute value extraction circuit 33 is larger than the threshold value A or smaller than the threshold value B, it is determined that the signal is other than block noise, and is output from the target pixel difference absolute value extraction circuit 33. Since the processed value is not sent to the subsequent stage, it is possible to prevent a false detection that the original video signal is detected as a vertical block boundary (block vertical boundary) in which block noise occurs. The resulting block vertical boundary can be accurately detected.

  Note that the present invention is not limited to the above embodiment, and for example, a computer program that causes a computer to execute the configuration of the block noise detection device shown in FIGS. 1, 2, 7, and 8 is also included in the present invention. It is included. This computer program may be recorded on a recording medium and taken into the computer, or distributed via a communication network and taken into the computer.

It is a block diagram of 1st Embodiment of this invention apparatus. It is a block diagram of 2nd Embodiment of this invention apparatus. It is a block diagram of an example of a vertical low-pass filter. It is a figure which shows an example of the pixel in the horizontal direction predetermined period, and its level. It is operation | movement explanatory drawing of the adjacent pixel difference absolute value extraction circuit in FIG. It is a figure explaining the operation | movement of the average value calculation circuit by a block horizontal pixel period in FIG. 1, and a block horizontal pixel periodicity detection circuit. It is a block diagram of 3rd Embodiment of this invention apparatus. It is a block diagram of 4th Embodiment of this invention apparatus. It is a block diagram of an example of a horizontal low-pass filter. It is a figure which shows an example of the pixel in the vertical direction predetermined period, and its level. It is a figure explaining the operation | movement of the average value calculation circuit by a block vertical pixel period in FIG. 7, and a block vertical pixel periodicity detection circuit.

Explanation of symbols

11 Vertical low pass filter (LPF)
12 An adjacent pixel difference absolute value extraction circuit within a predetermined period in the horizontal direction 13, 33 An attention pixel difference absolute value extraction circuit based on ranking of adjacent pixel difference values 14 An average value calculation circuit based on block horizontal pixel periods 15 A block horizontal pixel periodicity detection circuit 16 block horizontal boundary detection circuit 17, 37 attention pixel difference absolute value determination circuit with two thresholds 31 horizontal low pass filter (LPF)
32 An adjacent pixel difference absolute value extraction circuit within a predetermined period in the vertical direction 34 An average value calculation circuit based on a block vertical pixel period 35 A block vertical pixel periodicity detection circuit 36 A block vertical boundary detection circuit

Claims (6)

When block noise occurs in the decoded video signal when the encoded video signal obtained by compressing and encoding the video signal in units of blocks including a plurality of pixels is decoded in the block unit, the block noise is generated. In a block noise detection device for detecting a block boundary,
For each of a plurality of pixels in the horizontal period of the decoded video signal, an extraction means for calculating a difference absolute value with an adjacent pixel and extracting an adjacent pixel difference absolute value at a pixel position corresponding to each of the plurality of pixels;
The adjacent pixel difference absolute values at m pixel positions arranged in a time series within a predetermined period in the horizontal direction determined by the number of horizontal pixels m (m is a natural number of 2 or more) in block units are ranked in descending order. And at least one predetermined predetermined order including at least the largest order of the adjacent pixel difference absolute value at the target pixel position which is one pixel position set in advance in the horizontal direction predetermined period. When there is a process, the absolute value of the adjacent pixel difference at the target pixel position is output as a representative value for the predetermined period in the horizontal direction, and a numerical value 0 is output as the representative value when the rank is not the predetermined rank. The horizontal predetermined period is shifted by one pixel in the horizontal direction, and the representative value in each horizontal predetermined period is determined for each pixel position in the horizontal period. And ranking means for outputting a value,
The representative value output for each pixel position in the horizontal period is received as an input from the ranking unit, and the horizontal period is divided into predetermined horizontal periods, and m pieces are arranged in the horizontal predetermined period. An average for calculating an average value of representative values of pixel positions in the same order within each predetermined horizontal direction period when the pixel positions are the first to mth pixel positions in order from the front in time series. A value calculating means;
Of the average values at the first to m-th pixel positions calculated by the average value calculation means, when the largest average value is equal to or greater than a predetermined value determined based on the second largest average value The pixel position corresponding to the largest average value is extracted, and all the pixel positions appearing in the period unit of the horizontal predetermined period corresponding to the extracted pixel position in the horizontal period are included in the block boundary. Detecting means for detecting the pixel position;
At the pixel position where the block boundary is detected by the detection means, it is determined whether or not the fixed value is output from the ranking means. If the fixed value is not output, the pixel position is determined as block noise. And a block boundary detecting means for detecting the block boundary as a horizontal block boundary in which the noise occurs.   It is determined whether or not the representative value output from the ranking unit satisfies a condition that a preset value is different between a first threshold value and a second threshold value, and from the ranking unit When the output representative value satisfies the condition, the output value is output to the average value calculation means and the horizontal boundary detection means. When the output does not satisfy the condition, the output is output from the ranking means. The block noise detection apparatus according to claim 1, further comprising a determination unit that blocks output of the representative value to the average value calculation unit and the block boundary detection unit. When block noise occurs in the decoded video signal when the encoded video signal obtained by compressing and encoding the video signal in units of blocks including a plurality of pixels is decoded in the block unit, the block noise is generated. In a block noise detection device for detecting a block boundary,
For each of a plurality of pixels in the vertical period of the decoded video signal, an extraction unit that calculates a difference absolute value with an adjacent pixel and extracts an adjacent pixel difference absolute value at a pixel position corresponding to each of the plurality of pixels;
The adjacent pixel difference absolute values at m pixel positions arranged in time series within a predetermined period in the vertical direction determined by the number m of vertical pixels in the block unit (m is a natural number of 2 or more) are ranked in descending order. In addition, the adjacent pixel difference absolute value rank at the target pixel position, which is one pixel position set in advance within the predetermined period in the vertical direction, is at least one predetermined predetermined order including at least the highest order. When there is a process, the absolute value of the adjacent pixel difference at the target pixel position is output as a representative value in the predetermined period in the vertical direction, and a numerical value 0 is output as the representative value when the rank is not the predetermined rank The predetermined period in the vertical direction is shifted by one pixel in the vertical direction, and the representative value in each predetermined period in the vertical direction is calculated for each pixel position in the vertical period. And ranking means for outputting a value,
The representative value output for each pixel position in the vertical period from the ranking unit is received as an input, the vertical period is divided into predetermined vertical periods, and m pieces are arranged in the vertical predetermined periods. An average for calculating an average value of representative values of pixel positions in the same order within each predetermined vertical direction period when the pixel positions are the first to mth pixel positions in order from the front in time series. A value calculating means;
Of the average values at the first to m-th pixel positions calculated by the average value calculation means, when the largest average value is equal to or greater than a predetermined value determined based on the second largest average value The pixel position corresponding to the largest average value is extracted, and all the pixel positions appearing in the period unit of the predetermined period in the vertical direction corresponding to the extracted pixel position in the vertical period have the block boundary. Detecting means for detecting the pixel position;
At the pixel position where the block boundary is detected by the detection means, it is determined whether or not the fixed value is output from the ranking means. If the fixed value is not output, the pixel position is determined as block noise. And a block boundary detecting means for detecting the block boundary as a vertical block boundary in which the noise occurs.   It is determined whether or not the representative value output from the ranking unit satisfies a condition that a preset value is different between a first threshold value and a second threshold value, and from the ranking unit When the output representative value satisfies the condition, the output value is output to the average value calculation means and the vertical boundary detection means, and when the condition is not satisfied, it is output from the ranking means. 4. The block noise detection apparatus according to claim 3, further comprising a determination unit that blocks output of the representative value to the average value calculation unit and the block boundary detection unit. When block noise occurs in the decoded video signal when the encoded video signal obtained by compressing and encoding the video signal in units of blocks including a plurality of pixels is decoded in the block unit, the block noise is generated. A block noise detection program for detecting horizontal block boundaries by a computer,
The computer,
The block noise detection program according to claim 1, wherein the block noise detection program causes the extraction unit, the ranking unit, the average value calculation unit, the detection unit, and the block boundary detection unit to function. When block noise occurs in the decoded video signal when the encoded video signal obtained by compressing and encoding the video signal in units of blocks including a plurality of pixels is decoded in the block unit, the block noise is generated. A block noise detection program for causing a computer to detect a vertical block boundary,
The computer,
4. A block noise detection program that functions as the extraction means, the ranking means, the average value calculation means, the detection means, and the block boundary detection means according to claim 3.

Images