JP2003087823A - Transmission image quality monitor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a transmission image quality monitor for extracting a feature quantity from a transmitter side and a receiver side to supervise a remote place and capable of creating a valid block to apply simple orthogonal transform suitable for a monitor configuration. SOLUTION: Block division sections 41, 51 of first and second feature quantity extract sections 40, 50 form a square comprising 2's power number of pixels for a block and configure blocks by filling a predetermined value to blocks including outside of valid images of an image. Orthogonal transform sections 42, 52 apply orthogonal transform to each block and coefficient extract sections 43, 53 extract coefficients of the same position among the orthogonal transform coefficients. An MSE (Mean Square Error) estimate section 81 calculates an MSE estimate value on the basis of the coefficient received from the coefficient extract sections 43, 53. As for the blocks in which the prescribed value is filled, the MSE is estimated on the basis of a corrected MSE (=S/(S-K).MSE), where S is an area of block and K is an area filled by the prescribed value in the block area S.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a transmission image quality monitoring device, and more particularly, to a transmission image quality at each processing point in a system (hereinafter referred to as a video transmission chain) in which a plurality of transmission processing devices are arranged in series in a transmission path. The present invention relates to an automatic remote monitoring device for transmission image quality in video transmission, which is suitable for collectively remote monitoring in the central area.

[0002]

2. Description of the Related Art The present invention was invented by the present inventors as one of conventional remote automatic monitoring / evaluating devices for image quality by extracting feature quantities in a video transmission chain.
There is a patent application as No. 7226 "Transmission image quality monitoring device".

According to the invention of this patent application, an image is divided into blocks of a certain size, each divided block is orthogonally transformed, its transform coefficient is extracted, and the extracted transform coefficient is transmitted to a central monitoring room. This is to evaluate the transmission image quality. In implementing a device for orthogonally transforming a block obtained by dividing an image, application of Walsh Hadamard transform, Fourier transform, discrete cosine transform, or the like can be considered as the orthogonal transform.

It has been conventionally known that a high speed calculation method is preferably used in order to suppress the hardware scale of these conversions. Then, as a constraint to use the high-speed calculation method, the block size needs to be a power of 2.
Also, the larger the conversion block size, the larger the hardware scale.

Therefore, the smaller the block size, the better, and the size constraint is 2
It is desirable to be a power of.

[0006]

However, the following two problems arise here.

(1) Standard television (SDTV) and HDTV
Then, the screen size is not a power of 2. That is, the standard television has, for example, 720 (pixels) × 486 (columns), and the HDTV has, for example, 1920 (pixels) × 1080.
(Column). Therefore, a block of a size represented by a power of 2 can be used for standard televisions, HDTVs, etc.
How to apply it is practically important.

(2) If the block size is small, the number of blocks for covering the entire screen increases. Then, the extraction conversion coefficient to be transmitted to the central monitoring room is inevitably increased, and it becomes difficult to apply the monitoring device according to the present invention when only a thin monitoring line is provided.

The present invention has been made in view of the above-mentioned prior art, and an object thereof is to implement a transmission image quality monitoring apparatus for remote monitoring by extracting a feature amount from a transmitting side and a receiving side. An object of the present invention is to provide an apparatus capable of creating a block effective for applying a suitable simple orthogonal transform.

[0010]

In order to achieve the above-mentioned object, the present invention is a transmission image quality monitoring device for monitoring the transmission image quality on a video transmission line consisting of a cascade connection of a plurality of transmission devices. At a plurality of points on the transmission path, a feature amount extraction means for extracting a feature amount of video image quality, and MSE supplied with data from the feature amount extraction means via a low-speed line.
MSE estimation means for estimating (mean squared error), wherein the feature amount extraction means divides the input image into square blocks of powers of 2, and an orthogonal transform for orthogonally transforming the image in the block. The block dividing means comprises a transforming means and a frequency component value extracting means for extracting an arbitrary frequency component value of the orthogonal transformation. The block dividing means fills a block including a portion outside the effective screen of the image with a predetermined value, and
The first feature is that the blocks are configured as square blocks of powers of.

According to this feature, with respect to the block including the outside of the effective screen of the image, the outside of the effective screen is filled with a certain value, so that the orthogonal transform of the power of 2 for the screen of an arbitrary size. Can be applied. Therefore, the transmission image quality monitoring device can be simplified.

Further, according to the present invention, the feature amount extracting means is
Block dividing means for sampling distant pixels of the input video image and dividing it into square blocks of powers of two, orthogonal transformation means for orthogonally transforming the video image in the block, and frequency for extracting an arbitrary frequency component value of the orthogonal transformation. The second feature is that it is composed of a component value extracting means.

According to this feature, when the image is divided into blocks, each pixel is formed by sampling not the adjacent pixels but the pixels that are apart from each other, so that a wide range on the image plane can be formed by a relatively small block. Be able to cover. Therefore, the transmission image quality monitoring device can be simplified.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described in detail below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of an embodiment of the present invention applied to a video transmission chain which is a system in which a plurality of transmission processing devices are connected in cascade to a transmission line. It shows a part of the configuration.

In FIG. 1, it is assumed that the transmission image is transmitted from the transmission side 2 to the reception side 3 via the transmission line 1. The first feature amount extraction unit 40 includes, for example, a block division unit 41, a spread spectrum / orthogonal transformation unit 42, and a coefficient extraction unit 43, and extracts the feature amount from the transmission image on the transmission side 2. The second feature amount extraction unit 50 has the same configuration as the first feature amount extraction unit 40, and includes a block division unit 51, a spread spectrum / orthogonal transformation unit 52, and a coefficient extraction unit 53. And
The feature amount is extracted from the received image on the receiving side 3.

The feature quantities extracted by the first and second feature quantity extracting units 40 and 50 are sent to the central monitoring room 8 through low speed lines 6 and 7 such as telephone lines and LAN lines. In the central monitoring room 8, the MSE estimation unit 8 uses these feature quantities.
1. Estimate MSE (mean squared error) by 1.

In the first feature quantity extraction section 40, the block division section 41 divides the input image (transmission image) into blocks of a predetermined size, and the spread spectrum / orthogonal transformation section 42 spreads the spectrum / orthogonally. After conversion, the coefficient extracting unit 43 extracts an appropriate coefficient from the coefficients obtained by the orthogonal conversion. The second feature quantity extraction unit 5
Since the operation of 0 is the same as that of the first feature amount extraction unit 40,
The description is omitted.

As described above, in the transmission image quality monitoring apparatus of this embodiment, the functions of the spread spectrum / orthogonal conversion units 42 and 52 are important factors. When the orthogonal transform unit in the spread spectrum / orthogonal transform units 42 and 52 is implemented as a device,
In general, well known fast algorithms are often used. This is because it is possible to easily execute the orthogonal transform calculation, which is very large in scale in the general method. For this fast algorithm, for example, RJ Clarke,
“Transform coding of images”, Academic

Press, Microelectronics and Signal
Processing Series, 1985, pp. 291-335. However, the fast algorithm has a limitation on the transform block size. That is, the size is 2 for both vertical and horizontal.
It must be a power of.

In general, the orthogonal transform requires an enormous amount of calculation as the block size increases, and therefore it is important to reduce the size to a device size.

Therefore, the present inventor has set the block size to 2
In order to estimate the image quality as accurately as possible while keeping the exponentiation of power and keeping the size small, a virtual pixel value (dummy bit) that does not exist in the original screen is inserted into the block to configure the block, The coefficient values obtained by orthogonally transforming the screen are extracted by sampling.

Next, this embodiment will be described in detail with reference to FIG. For example, in a standard television (hereinafter referred to as SDTV), one field consists of 720 pixels in the horizontal direction × 243 lines in the vertical direction, as shown in the figure. At this time, the block size is, for example, 128 × 128.
Assuming (2 ⁷ × 2 ⁷ ), in order to cover one field, 6 blocks (eg, B _{11 to} B ₆₁ ) in the horizontal direction and 2 blocks (eg, B _{1 1 to} B ₁₂ ) in the vertical direction are covered. Need to be lined up.

At this time, when trying to align the screen and the upper left corner of the block, the blocks in the rightmost column and the bottom row (B ₆₁ , B
_{12 to} B ₆₂ ), there will be a portion not included in the screen. That is, 128 × 6 = 768, 1
Since 28 × 2 = 256, 4 is assigned to the rightmost block.
8 × 128 pixels, 128 × 13 pixels are not included in the bottom row block, that is, a blank of pixels occurs.

Therefore, in the block division units 41 and 51,
The blank portion of the rightmost column and lowermost row block, arbitrary constant value, for example, the dummy bit "128" (= half of ²⁸⁾
Is embedded virtually, and the spread spectrum / orthogonal transformation unit 4
I will send it to 2, 52. The spread spectrum / orthogonal transform units 42 and 52 perform spread spectrum and orthogonal transform on the transmitted block, and the coefficient extraction units 43 and 53 perform arbitrary one of the coefficients of the orthogonal transform f _i , f. _i '
(However, the same coefficient at the same position in the same block) is extracted for each block one or more, and the MS of the central monitoring room 8
It is sent to the E estimation unit 81.

The operation of the MSE estimation unit 81 will be described with reference to the flowchart of FIG. First, in step S1, the MSE estimating unit 81 acquires the coefficients f _i and f _i ′ from the coefficient extracting units 43 and 53. In step S2, the coefficient f
It is determined whether or not _i and f _i 'are coefficients of the block in which the dummy bit is embedded. If the block is not embedded with the dummy bit, the process proceeds to step S3, which is an application by the present inventor. Japanese Patent Application No. 2001-11
MSE is obtained by the method described in No. 7226 "Transmission image quality monitoring device". That is, MSE is | f _i −
It can be obtained from f _i '| ² . Incidentally, in the case of extracting the orthogonal transformation coefficients of the plurality per one block by the coefficient extraction unit 43, 53, MSE ^{_{is, | f i -f i '|}} 2
It can be calculated from the average of.

On the other hand, in the case of the block in which the dummy bit is embedded (the judgment in step S2 is affirmative), the process proceeds to step S4, where MSE is S / (SK) .MSE.
And correct it. Here, S is the area of the block, and K is the area of the block area S filled with dummy bits.

The reason for correcting the MSE as described above is as follows. The image data sent from the transmission side 2 to the reception side through the transmission path 1 are all images in the effective screen area,
Areas in which dummy bits are virtually filled are not transmitted. Therefore, the MSE that is actually desired is all the image quality deterioration of the image in the effective screen area. However, MSE
The MSE calculated by the estimation unit 81 is the block division unit 4 for the block in which the dummy bit is embedded.
This is a value affected by the dummy bits embedded with 1,51. Moreover, since the dummy bit does not pass through the transmission line 1 and the like, the image quality is not deteriorated at all.

Therefore, the aforementioned Japanese Patent Application No. 2001-117226.
Considering that the MSE of the corresponding block estimated by the method shown in No. “Transmission image quality monitoring device” is an average value per pixel when the entire block is uniformly deteriorated, The estimated value of the original MSE of the effective screen portion in the block including the area in which the bits are virtually filled is
It becomes S / (SK) MSE.

Next, in step S5, it is judged whether or not the MSEs of all the blocks of one image have been obtained.
If this determination is negative, the process proceeds to steps S6 and S1,
The coefficient of the next block is input to the MSE estimation unit 8a.
When the above processing is continued and the processing of one image is completed (Yes in step S5), the process proceeds to step S7, and it is determined whether the processing of the entire image sequence is completed. If this judgment is negative, step S8,
Proceeding to S1, the MSE estimating unit 8a acquires the coefficient of the block of the next image.

When the above processing is continued and the processing of the entire image sequence is completed (determination in step S7 is affirmative)
Proceeding to step S9, the calculation of the estimated value of MSE of the entire image sequence is performed as follows. That is, MSEs of blocks (total of n blocks) in which all blocks are occupied by valid screens are MS ₁ , MS ₂ , ..., MS _n , and the right part of the blocks is outside the screen (area KR) The estimated values of MSE (total of m pieces) are MR ₁ , MR ₂ , ...
, MR _m , MB estimated value of MSE of block (total of p blocks) where lower part of block is outside screen (area KB)
₁ , MB ₂ , ..., MB _p , M of a block (total of q pieces) whose lower right portion inside the block is outside the screen (area KBR)
If the SE estimates are MBR ₁ , MBR ₂ , ..., MBR _q , the MSE estimate for the entire image sequence is
It becomes like the formula (1). In addition, the above MSE
← S / (SK) MSE is also taken into consideration.

As a special case, one screen is exactly 2
When expressed by a power of, the above-mentioned KR, KB, KB
Since R, m, p, and q are all 0, the above-mentioned Japanese Patent Application No. 2001
This is in agreement with the formula expressed by No. 117226 “Transmission image quality monitoring device”.

Next, a second embodiment of the present invention will be described with reference to FIG. 4, the same or equivalent parts as those in FIG. 1 are designated by the same reference numerals. In the present embodiment, the first feature quantity extraction unit 60 uses the sub-sampling block division unit 6
2, a spread spectrum / orthogonal transformation unit 63, and a coefficient extraction unit 64, and a second feature amount extraction unit 70.
Is composed of a sub-sampling block division unit 72, a spread spectrum / orthogonal transformation unit 73, and a coefficient extraction unit 74, which have the same configuration as the first feature quantity extraction unit 60. The band limiting filters 61 and 71 may be omitted because they are not necessarily required configurations, but if these configurations are adopted, it is possible to detect the influence of local breakdown of pixels that are not subsampled. .

Next, the operation of this embodiment will be described. Similar to the case of the first embodiment, considering a case where the screen is a standard television size (1 field is 720 pixels in the horizontal direction × 243 lines in the vertical direction), for example, a size of 8 × 8 pixels which is easy to implement If the screen is to be divided into blocks, the number of blocks becomes (720/8) × {round up the decimal point of the integer value of (243/8)} = 2790 blocks. If only one coefficient is extracted from one block, if the coefficient representation bit precision is 10 bits, then 1
The amount of extraction coefficient information per field is 27.9 kbits (= 2790 × 10), which is 1.674 in 1 second.
It becomes Mbps (= 27.9 × 60). This is an excessively large amount of information to be sent over the low-speed lines 6 and 7 for monitoring such as a telephone line and a LAN line, so it is necessary to reduce the amount of data by some method. However, if data is to be extracted from only a part of the screen in order to reduce the amount of data, there is a risk of overlooking a local image failure peculiar to a digital transmission error.

Therefore, in the present embodiment, the pixels that are adjacent to each other are not collected into one block, but the pixels that are apart from each other are collected.

In FIG. 4, the transmitted image and the received image are divided into sub-sampling block dividing units 62 and 7, respectively.
It is divided into two blocks. In the sub-sampling block division units 62 and 72, both in the horizontal direction and the vertical direction,
Sub-sampling is performed at a ratio of 1 pixel to 8 pixels, and the sub-sampled 8 × 8 pixels form one block. Therefore, the number of blocks created by the sub-sampling block division units 62 and 72 is 1/64 of the above 2790.
And 44. Then, the amount of data sent on the low speed lines 6 and 7 for monitoring is 26.4 kbps per second.
s, and the order can be transmitted even on the low speed lines 6 and 7.

The sub-sampling of 1 pixel out of 8 pixels is appropriate for the following reason. That is, MPEG-2, which is usually used for moving image compression transmission, is 8 × 8.
Since the processing is performed in pixel block units, the influence (failure) on the image due to a transmission path error or the like often appears in this block unit. Therefore, 1 in a block of 8 × 8 pixels
By subsampling each one individually, it means that the minimum necessary needs for monitoring are met.

Next, the blocks obtained by the sub-sampling block division units 62 and 72 are spectrum-spread by the spread spectrum / orthogonal transform units 63 and 73 to average pixel data and then orthogonally transformed, and the coefficient extraction unit 64, At 74, the coefficient at the same position is extracted. The coefficient extracting units 64 and 74
The coefficients f _i and f _i ′ extracted in step 1 are sent to the MSE estimation unit 91 in the central monitoring room 8 via the low speed lines 6 and 7. M
The SE estimation unit 91 uses the method described in Japanese Patent Application No. 2001-117226 “Transmission image quality monitoring device”
Estimate E.

[0038]

As is apparent from the above description, according to the present invention, by embedding a predetermined constant value in a block including the outside of the effective screen of an image, a screen of an arbitrary size can be set to 2 It becomes possible to apply the orthogonal transform of the power-size orthogonal transform. According to the present invention, this makes it possible to use high-speed orthogonal transform, which is relatively easy to implement, and simplifies the transmission monitoring device.

Further, according to the present invention, when the image is divided into blocks, not the pixels but the pixels which are distant from each other are sampled so as to configure each part lock. You will be able to monitor the entire screen with small blocks. That is, the transmission monitoring device can be downsized.

[Brief description of drawings]

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

FIG. 2 is a conceptual diagram of block division according to the present embodiment.

FIG. 3 is a flowchart illustrating an operation of an MSE estimation unit of this embodiment.

FIG. 4 is a block diagram showing a schematic configuration of a second embodiment of the present invention.

[Explanation of symbols]

1 ... Transmission path, 2 ... Transmission side, 3 ... Reception side, 6,7 ... Low speed line, 8 ... Central monitoring room, 40,60 ... First feature quantity extraction unit , 50, 70 ... Second feature quantity extraction unit, 41, 51 ...
Block division unit, 42, 52 ... Spread spectrum / orthogonal transformation unit, 43, 53 ... Coefficient extraction unit, 62, 72 ... Sub sampling block division unit, 63, 73 ... Spread spectrum / Orthogonal transformation Part, 64, 74 ... Coefficient extraction part, 8
1, 91 ... MSE estimation unit.

   ─────────────────────────────────────────────────── ─── Continued front page    F-term (reference) 5C061 BB13 CC03                 5J064 AA01 BA16 BB03 BC01 BC18                       BD02

Claims (4)

[Claims] 1. A transmission image quality monitoring device for monitoring the transmission image quality on a video transmission line composed of a plurality of cascaded transmission devices, wherein the feature amount extracts the feature amount of the video image quality at a plurality of points on the video transmission line. And an MSE estimating unit for estimating the MSE (mean square error) supplied from the feature amount extracting unit via the low-speed line. The block dividing means includes block dividing means for dividing into power square blocks, orthogonal transform means for orthogonally transforming an image in the block, and frequency component value extracting means for extracting an arbitrary frequency component value of the orthogonal transform. A transmission image quality monitoring apparatus is characterized in that a block including a portion outside the effective screen of an image is filled with a predetermined value to form the square block of the power of 2. 2. The MSE estimation means, for a block in which the predetermined value is embedded, fills the area of the block with S and the predetermined value of the block area S. 2. The transmission image quality monitoring device according to claim 1, wherein when the area is K, MSE is corrected to obtain S / (SK) .MSE. 3. A transmission image quality monitoring device for monitoring the transmission image quality on a video transmission line consisting of a plurality of cascaded transmission devices, wherein the feature amount extracts the image quality feature amount at a plurality of points on the video transmission line. It comprises an extracting means and an MSE estimating means for estimating an MSE (mean square error) supplied from the characteristic amount extracting means through a low speed line, wherein the characteristic amount extracting means separates the input image from each other. Block division means for sampling pixels to divide them into square blocks of powers of two, orthogonal transformation means for orthogonally transforming an image in the block, and frequency component value extraction means for extracting an arbitrary frequency component value of the orthogonal transformation. A transmission image quality monitoring device comprising: 4. The transmission image quality monitoring device according to claim 3, wherein the block dividing means samples pixels from 8 × 8 pixels of the input image.