JP2008124901A - Video signal processing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect a plurality of block widths in simple configuration for a signal which is encoded and decoded by block units and with respect to the position or the width of a block boundary is unknown. <P>SOLUTION: A video signal processing apparatus includes a block boundary determining section 1 which performs operation with pixels around each pixel of an inputted image signal for each pixel, to determine whether the position of the pixel is a block boundary candidate; a histogram creating section 2, which samples a result of the determination in a term that is a common multiple of a plurality of block widths that are desired to be determined and takes a histogram with respect to an object for which the remainder of an address at a pixel position with respect to the common multiplier is smaller than a predetermined constant; and a block width/position determining section 4 which determines the width and the position of the block from the histogram. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an apparatus and a method for detecting block noise that occurs when an image is block-encoded / decoded.

  In a system in which an image such as MPEG is encoded and transmitted in block units, block-like distortion (block noise) may occur after decoding. In order to reduce such block noise, there is a method of using a smoothing filter at the block boundary part. In this case, the block boundary needs to be known.

  When processing can be performed subsequently after decoding, it is possible to identify and process block boundaries, but when processing video signals that have been decoded by other devices, The position becomes unknown.

  In addition, the block unit processing is usually performed in units of eight pixels. However, in the case of a signal decoded by another device, the pixel number conversion processing may be performed in addition to the decoding, and thus the block is not necessarily used. May not be 8 pixels wide.

  Therefore, a method for detecting a block boundary from a signal whose block boundary position and width are unknown is required.

  Patent Document 1 discloses a method of taking a histogram of pixel addresses determined to have a block boundary in order to detect the position of a block having a width of 8 pixels, using the remainder of the pixel width, and specifying the position.

  Patent Document 2 discloses the following technique as a method corresponding to a plurality of block widths.

  First, when detecting a horizontal block boundary, a column difference array is obtained by averaging the absolute values of the difference values between adjacent columns in the vertical direction, and this column difference array is averaged by the block width. Thus, a block average column difference array is generated.

  In the block average column difference array, the ratio of the peak and the average value other than the peak is taken. If the ratio is equal to or greater than the threshold value, it is determined that the block boundary is at the peak position.

If multiple block widths are assumed, repeat the block width averaging operation for multiple block widths to obtain multiple peak-to-peak average ratios, and use the maximum block width as the image block width. adopt.
JP 2002-354480 A JP 2002-077629 A

  However, in the above-described conventional technology, although the position can be specified in Patent Document 1, only a certain fixed block width is assumed, and in Patent Document 2, although a plurality of block widths are supported. First, in order to obtain a column difference array, it is necessary to secure a memory corresponding to the width of the image, and it is necessary to perform an averaging operation for each of a plurality of block widths.

  The present invention samples a block boundary determination result every common multiple of a plurality of block widths to be obtained, and uses a histogram of the remainder for the common multiple of the pixel address to detect a plurality of block widths with a simple configuration. Objective.

In order to solve the above-described problems, the present invention provides:
Determining means for determining whether or not the pixel position is a block boundary candidate by calculation with the surrounding pixels for each pixel of the input image signal;
Histogram creation means for sampling the determination result at a period of a common multiple of a plurality of block widths to be obtained, and taking a histogram with respect to an address of the pixel position with respect to a thing whose remainder with respect to the common multiple is smaller than a predetermined constant;
It has a judging means for obtaining the width and position of the block from the histogram.

  According to the present invention, the block boundary position can be detected with a simple circuit configuration even when a plurality of block widths are assumed.

(Example 1)
Embodiments of the present invention will be described below with reference to the drawings.

  In this embodiment, the horizontal block width is detected.

  As the image size, when the image is not enlarged and the block width is 8 pixels, the original image whose horizontal width is 1440 pixels is enlarged to 1920 pixels, and the block width is 32/3 pixels. In the case where the width of the original image is 1280 pixels, the original image is enlarged to 1920 pixels, and an example corresponding to three block widths when the block width is 12 pixels is shown.

  FIG. 1 is a block diagram showing a configuration of an embodiment of the present invention.

  1 is a block boundary determination unit, 2 is a histogram generation unit, 3 is a counter, and 4 is a block width / position determination unit.

The image signal is first input to the block boundary determination unit 1.
The block boundary determination unit 1 performs block boundary determination from the difference value between the data of four adjacent pixels.

Specifically, as shown in FIG. 2, four adjacent pixels are P1, P2, P3, and P4, and the values of these luminance signals are L (P1), L (P2), L (P3), and L (P4). )
| L (P1) -L (P2) | <th1 and | L (P3) -L (P4) | <th1 and
th1 <| L (P2) -L (P3) | <th2
When the condition is satisfied, 1 is output assuming that there is a block boundary between P2 and P3, and 0 is output if the condition is not satisfied. th1 and th2 are predetermined thresholds satisfying th1 <th2.

  This determination condition is that when there is a block boundary between P2 and P3 and block noise is generated, there is a certain luminance difference between pixels on the block boundary, and there is a luminance difference between pixels in the same block. It is an example of a boundary determination condition based on the property of being small, and other boundary determination methods may be used.

  The determination result output from the block boundary determination unit 1 is input to the histogram generation unit 2. The histogram generation unit 2 generates a histogram based on the block boundary determination result and the output from the counter 3.

  The counter 3 outputs the address of the pixel P3 in FIG.

  The histogram generation unit 2 generates a histogram according to the flowchart of FIG. 3 based on the block boundary determination result and the output from the counter 3.

  First, the block boundary determination result and the pixel address from the counter are input (step S1). When the block boundary determination result is 1, and it is determined whether the remainder obtained by dividing the input pixel address by 96 is less than 43. If the condition is satisfied, the processing is continued. Then, the process proceeds to the next pixel (step S2). If the processing is continued, 1 is added to the remainder of 96 pixel addresses (step S3).

  The above determination is performed on the pixels in one screen to create a histogram (step S4).

  The generated histogram is input to the block width / position determination unit 4 for determination processing.

  The block width / position determination unit 4 detects a peak in the histogram and obtains an interval between the peaks.

  The peak is compared in order from the maximum value in the histogram, and it is determined that the peak is the point where the difference from the next value is less than or equal to the threshold value.

  When the peak is obtained in this way, the peak interval is obtained from the difference between the peaks.

  The block width is determined from the peak interval and the appearance pattern thus obtained.

  If the peak is every 8 pixels as shown in FIG. 4A, it is determined that the block is 8 pixel wide.

  If it is a peak every 12 pixels as shown in FIG. 4B, it is determined as a 12 pixel wide block.

  If there is a pattern of 11 pixels, 10 pixels, and 11 pixels as shown in FIG. In the case of a 32/3 pixel width block, the determination is similarly made with a pattern of 11 pixels, 11 pixels, 10 pixels, or any interval of 10 pixels, 11 pixels, and 11 pixels.

  Once the block width is determined, the position can be detected at the peak position.

  Here, the reason why the block width can be determined from the residual histogram of 43 pixel addresses will be described.

  If the block width is X pixel width, one peak appears if a histogram of block boundary determination results is taken at intervals of X pixels. If a histogram is taken at intervals of n times the X pixels, n peaks appear.

  When there are multiple block widths, if you take a histogram of block boundary judgment results with a period that is a common multiple of those pixel widths, a peak corresponding to that pixel width will appear no matter what pixel width the input is Will do.

  Therefore, it is possible to detect the peak from the histogram of the block boundary determination result taken at a period that is a common multiple of a plurality of block widths and determine the width and position of the block.

  In the case of the present embodiment, since there are three of 8 pixel width, 32/3 pixel width, and 12 pixel width, it is sufficient to take a histogram with a 96 pixel width which is the least common multiple.

  Here, in order to determine the width and position of the block, it is necessary to histogram the block boundary determination result with a common multiple of a plurality of block widths, but the width of the common multiple is not necessary as the width of the histogram.

  For example, when the width is 8 pixels, the width of the histogram always includes two peaks. When the width is 16 pixels, two peaks are always included.

  In the case of a 12-pixel width, two peaks are always included in a 24-pixel-wide histogram.

  In the case of 32/3 pixel width, since the pixel width is not an integer, the peak interval is 11 pixels, 11 pixels, 10 pixels, or 11 pixels, 10 pixels, 11 pixels, or 10 pixels, 11 pixels, 11 Any pattern of pixels is repeated. For this reason, the width of the histogram in which one pattern is always included is necessary, and if there is a width of 43 pixels, one pattern is always included.

  Therefore, the maximum value of the minimum pixel width necessary to detect each block width, and in the case of this embodiment, a 43 pixel width histogram is taken to detect each block width and position. Can do.

  As described above, according to the present invention, block boundary positions can be detected corresponding to a plurality of block widths with a simple configuration.

  Although the present embodiment has been described with respect to horizontal processing, block boundaries can be detected by performing similar processing in the vertical direction.

  In this embodiment, the case where the block width is 8, 32/3, or 12 has been described. However, the present invention is not limited to this, and combinations of four or more block widths or different block widths are possible. Even in this case, the block boundary can be similarly detected by sampling the determination result at a cycle of a common multiple of a plurality of block widths to be obtained and taking a histogram with a remainder of the common multiple with respect to a pixel address.

It is a block diagram of the Example of this invention. It is a figure explaining a block boundary determination method. It is a flowchart explaining the process sequence of the histogram production | generation part 2 of this invention. It is a figure explaining the method of determining a block width from a histogram.

Explanation of symbols

1 Block boundary determination unit 2 Histogram generation unit 3 Counter 4 Block width / position determination unit

Claims (2)

An image processing apparatus that determines one of a plurality of block widths,
Determining means for determining whether or not the pixel position is a block boundary candidate by calculation with the surrounding pixels for each pixel of the input image signal;
Histogram creation means for sampling the determination result at a cycle of a common multiple of a plurality of block widths to be obtained, and taking a histogram with respect to an address of the pixel position with respect to a thing whose remainder with respect to the common multiple is smaller than a predetermined constant;
An image processing apparatus comprising: a determination unit that obtains the width and position of a block from the histogram.   2. The image processing according to claim 1, wherein the predetermined constant for taking the residual histogram is a maximum one of pixel widths necessary for identifying each of the plurality of block widths. apparatus.

Images