JPH07118809B2 - Image transmission method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to transmission of image information for a TV monitoring system, and more particularly to an image transmission method for narrow band transmission of only a changed portion when an image changes.

[Background technology]

TVs have traditionally been used in image change detection and transmission methods.
In order to transmit a moving image like a conference system, a system that transmits only a difference between image frames, that is, only a changed portion thereof has been put into practical use. That is, if the frame is updated at a frame cycle (1/60 second) such that 30 frames are transmitted per second by interlacing like a general NTSC system TV image, if it is called a real-time moving image, A raw image requires a transmission band of about 50 Mbits / sec when quantized to 6 bits per pixel, but when quantized an image frame to 256 × 256 pixels, it is about 11.8 M when transmitted at 30 frames per second.
The transmission bandwidth is bit / second, but 1 bit / pixel when compressing information without degrading image quality except redundancy due to pixel correlation (approximately 0.9) or image frame correlation (approximately 0.8). It seems that it requires a transmission bandwidth of 2 Mbit / sec. Here, in the video conference experimental system reported by Nippon Telegraph and Telephone Public Corporation, the transmission band is 1.5 Mbit / sec,
Also, in the conversational videotex, the transmission band is 8 Kbit / sec. By the way, since it is not INS at this time, 1.2Kbit /
The transmission band is from seconds to 9.6 Kbits / second. Therefore, 30 / (2M / 1.2K) = 0.02 frames / sec for transmission at 1.2 Kbits / sec without degrading the image quality of image frames, and the transmission rate of about 56 sec / frame is the limit.

In the TV surveillance system, a still image can be transmitted in 20 to 30 seconds in a transmission band of 9.6 Kbits / second, but it has been put into practical use. Is about 1/5 of the entire image (correlation between frames is 0.8), the frame period is 4 to 6 seconds / frame. If you use a 1.2Kbit / sec modem for low cost transmission, it takes 8 times as long as 9.6Kbit / sec and even if a still image is compressed to 2bit / image,
The transmission time for one image frame is about 130 seconds.

Therefore, the present inventors have proposed a method (Japanese Patent Application No. 59-130583) that uses an information compression method based on variable sample density predictive coding. The still image can be transmitted in about 60 seconds with the 1.2Kbit / sec transmission band by the intraframe compression coding method that can reproduce the main contours and give an overview of the image. The present inventors also proposed a method (patent application No. 59-18310) for transmitting image data only in the changing region by variable sampling density predictive coding. If
It can be transmitted at a frame cycle of about 12 seconds at 1.2 Kbits / second. By the way, when transmitting the image data of only the changed area, the area that has not changed is the still image received previously, so a 256 × 256 pixel still image is required at the time when a high quality image is required. There was a drawback that it took about 130 seconds when transmitting at 1.2 Kbits / second.

On the other hand, in the case of moving image transmission, the difference value between frames in the unchanged portion becomes small and the image quality improves when the frame cycle is repeated. It seems that it can be transmitted in about 60 seconds. However, conventional moving images have been studied at 1.2 Kbit / sec, and a method of compressing at a high rate so that the moving image can be transmitted and has practically no problem.

FIG. 5 shows a Japanese Patent Application No. 59-1831 which is the most basic example of the method of the present invention.
It shows the overall structure of No. 0. In the figure, first, a video signal captured by a camera 1 such as an ITV is separated into a sync signal and an image signal by a sync separation circuit 2, and the image signal is A /
It is converted into 8-bit data by the D converter 3, and is written into a designated frame memory among the two frame memories 6a6b by the address from the address counter 4 and the switch 5. The switch 5 is for alternately switching the frame memories 6a6b every time one frame is written to obtain the difference between the previous screen and the current screen. The absolute value conversion circuit 7 removes the sign from the difference between corresponding image data (8 bits) on both screens or squares the difference to convert the difference into an absolute value. The projection calculation circuit 8 cumulatively adds the absolute value of the difference of each pixel for each line parallel to each coordinate axis. As shown in FIG. 6, the X axis is on each line parallel to the Y axis. The total of the absolute value data of the difference (or it is divided by the number of pixels of one line) is projected, and the total of the data on each line parallel to the X axis is projected on the Y axis. The change area determination circuit 9 detects the minimum and maximum values X ₁ , X ₂ and Y ₁ , Y ₂ of X and Y corresponding to the line whose cumulative addition value exceeds the change detection setting level in FIG. , X ₁ ≦ X ≦ X ₂ and Y ₁ ≦ Y ≦ Y ₂ are determined. The image information in the change area is compression-encoded by the compression-encoding circuit 10 and narrow-band transmitted by the telephone line 15 via the transmission controller 11 and the modem 12. Further, the cumulative addition value of any line in the abnormality judgment area preset by the cumulative judgment circuit 13 is the sixth value.
When the abnormality detection set level shown in the figure is exceeded, an alarm signal is sent from the alarm generation circuit 14. On the receiving side, the transmission signal is received by the reception controller 17 via the modem 16, and in the case of an alarm signal, it is sent to the alarm notification circuit 18 to notify the abnormality by a buzzer or a lamp. In the case of image information, it is first sent to the change area setting circuit 19 and the change area (X ₁ , Y ₁ ) (X ₂ ,
Y ₂ ) is detected, the decompression decoding circuit 20 decompresses the information, and the information is written in the frame memory 21. The contents of the frame memory 21 are read by the address counter 22 and input to the D / A converter 23, and at the same time, the periodic signal from the synchronization generating circuit 24 is combined with this image signal by the synchronization combining circuit 25, and the result is monitored. Displayed on TV26.

FIG. 7A shows each concrete circuit of the compression encoding circuit 10 and FIG. 7B shows each concrete circuit of the decompression decoding circuit 20. The compression coding circuit 10 in the figure (a) is a variable sample density prediction coding circuit configured by combining a variable sample density coding circuit and a previous value prediction circuit, and is provided in a feedback loop outside the previous value prediction circuit 27. A variable sample density compression circuit 28 and a variable sample density decompression circuit 29 by the variable sample density method are inserted, and further, a previous value prediction circuit
Line buffer for 1 line in feedback loop inside 27
30 is provided, and the previous value prediction is performed using the line buffer 30 and the sampled value from the next line. The variable-sample-expansion decoding circuit 20 of FIG. 7B is a variable-sample-density prediction decoding circuit configured by combining a variable-sample-density decoding circuit and a previous-value prediction decoding circuit. The line buffer 33 is provided in the loop of the prediction circuit 32 for restoring the sample from the data expanded by the circuit 31.

FIGS. 8 (a) and 8 (b) illustrate an example of the variable sampling density coding method. The relationship between the sampling period and the sample value is defined by a triangle, and the triangle is moved as shown. Then, by transmitting the difference indicated by the arrow, the sampling interval is obtained from the difference value and the triangle on the receiving side, and the original waveform is restored. The smaller the variation of the sample value, the longer the sampling interval and the more the data is compressed.

FIG. 9 shows a state in which a previous value is predicted in the horizontal direction on a horizontal scanning screen, and prediction residuals arranged in the vertical direction are subjected to variable sample density coding in the vertical direction. The prediction value of each pixel is obtained by adding the previous value prediction residual, which has been subjected to variable sample density decoding in the loop, to the previous value held in the line buffer 30, and multiplying it by the prediction coefficient. Actually, the prediction coefficient is set to 1.0 and multiplication is not performed.

The present inventors have already proposed a method in which the above basic example is further improved. Next, this method will be described. FIG. 10 shows an example of the circuit configuration. FIG. 10A shows an inter-frame variable sample density predictive compression coding circuit provided on the transmission side.
34, and FIG. 14B shows an inter-frame variable sample density predictive decompression / decoding circuit 38 provided on the receiving side of the present embodiment. The inter-frame variable sample density predictive compression encoding circuit 34 is an intra-frame variable sample density predictive compression encoding circuit 39.
, A prediction frame buffer 35, and an inter-frame previous value prediction circuit 36, and a sample is stored in the current image frame buffer 37.
It is designed to be taken in via. If the hardware configuration enables real-time processing, the current image frame buffer 37 on the image input side becomes unnecessary. Since the processing of one frame requires a time on the order of seconds when processed by a microcomputer, the current image frame buffer 37 is provided as a frame buffer for one screen. The sample is a linear scale, and if it is a digital value of 6 bits to 8 bits and 256 × 256 pixels, the sampling period of 1 pixel is 200 nse c
There is. Thus, a sample of the current image input from the image pickup device (not shown) via the current image frame buffer 37 and prediction from the transmitted previous image written in the previous image prediction area 35b of the prediction frame buffer 35 The difference from the image sample enters the intra-frame variable sample density prediction compression circuit 39 in the feedback loop outside the inter-frame prior value prediction circuit 36, and the compressed code is transmitted to the reception side. This intra-frame variable sample density prediction compression circuit 39 has the same structure and operation as the compression encoding circuit 10 of FIG. 7 (a), and has the following features. In other words, as shown in FIG. 9, the two-dimensional redundancy is removed, and as shown in FIG. 8, the interval between samples can be doubled when the prediction residual becomes 1/2. Can be raised. In addition, the relationship between the prediction residual and the sampling interval has a one-to-one correspondence (called a quantization characteristic) in which the sampling interval becomes shorter as the prediction residual increases, and if only the prediction residual is transmitted. Since the receiving side can perform decoding based on the quantization characteristic, the amount of transmission information can be reduced. Furthermore, if the degree of expansion / contraction of the quantization characteristic in the time direction is controlled so that the sampling interval becomes longer when the prediction residuals are the same, the degree of compression can be increased.
Furthermore, the quantization in the amplitude direction of this quantization characteristic is nonlinearly quantized according to the distribution of the prediction residual so that the smaller the residual becomes, the denser it becomes. It is equivalent to quantizing a rough and small part densely, and if the shape of the quantization characteristic is shortened with respect to the time axis, it becomes close to DPCM. Moreover, as shown in FIG. 9, since the redundancy is removed two-dimensionally, the prediction residual can be reduced and the compression rate can be increased, and the prediction residual in the direction of the previous value prediction shown in FIG. Since it is calculated for each line, the main contour is not disturbed even when compressed to more than the redundancy and the fine details are not transmitted.Therefore, the contour of the image is more accurate than the original contour with less pixels. There is a feature that can be transmitted to. And "One Method of Variable Sampling Density Coding", IEICE Transactions '75 / 2 Vo
As described in l58 A No 2, P97 (1975), the variable sampling density coding method is particularly advantageous for a signal with a small change, and by applying it to the frame difference signal of the image signal, It has a characteristic of being able to perform a large band compression, and according to the three-dimensional prediction added between frames, even if the compression rate is higher than the redundancy, the main contour information can be transmitted.

In the feedback loop outside the intra-frame variable sample density predictive compression coding circuit 39, the variable sample density expansion circuit 29 expands (decodes) in the same way as the receiving side, and the outside of the inter-frame variable sample density predictive compression coding circuit 34. In the feedback loop, by adding with the above-described predicted value of the previous image from the feedback loop inside the inter-frame variable sample density predictive compression encoding circuit 34, it is expanded (decoded) in the same manner as the receiving side, and the predicted frame buffer is obtained. It is stored in the current image predicted value area 35a of 35. Therefore, since the redundancy is three-dimensionally removed by inter-frame and intra-frame composite prediction, and the number of transmission samples is reduced by the variable sample density coding, highly efficient compression is possible. Furthermore, since the transmitting side uses the same predicted image as the receiving side as the image of the previous frame, the transmission error is transmitted as the prediction residual, so the stationary part asymptotically becomes the current image by repeating the frame period. Since it approaches, it is possible to send an image with good image quality faster for a gentle change compared to the previous still image transfer, and it is possible to transfer a rough image in a short time without losing the main contour information even for a sudden change. become. In the quantization characteristics of variable sample density encoding, the smaller the prediction residual of the amplitude quantization level, the higher the compression rate and the better the image quality. Therefore, the inter-frame previous value prediction circuit 36 that matches the inter-frame correlation of the target monitoring image
It is desirable to set the prediction coefficient of. This also applies to the previous value prediction circuit 27 in the frame. In particular, when the line buffer 40 is provided in the feedback loop inside the inter-frame previous value prediction circuit 36 as shown in FIG. 10 (a), the predicted value of the predicted frame buffer 35 becomes the same as that on the receiving side. In this case, looking at the contents of the prediction frame buffer 35 on the monitor TV 26, the prediction coefficients of the inter-frame previous value prediction circuit 36 and the previous value prediction circuit 27 are set to 0.6 so that the transmission of a still image (one frame) becomes the optimum image. , 0.7, 0.8, 0.9, 1.0 and so on. As a result, an image with better image quality can be transmitted faster.

The inter-frame variable sample density predictive decompression / decoding circuit 38 on the receiving side shown in FIG. 10 (b) is the same as the decompressing / decoding circuit 20 shown in FIG. 7 (b). A prediction frame buffer 42 having a current image prediction region 42a and a previous image prediction region 42b, and an inter-frame previous value prediction circuit 43.
, And decodes the transmitted code to restore the sample. The prediction coefficients of the prediction circuits 32 and 43 can be selected similarly to the transmitting side.

FIG. 10 (c) shows another example of the inter-frame variable sample density predictive compression encoding circuit 34. In this example, the content of the prediction frame buffer 35 is multiplied by the prediction coefficient in the inter-frame previous value prediction circuit 36. It has become a thing.

[Object of the Invention]

The present invention has been made in view of the above-mentioned problems, and is limited to only a level at which it can be determined that the change between frames of an image is significant, even when the frame period is short and the amount of change is large. An object of the present invention is to provide an image transmission method capable of transmitting an outline in a short time, and at the same time, to provide an image transmission method capable of efficiently transmitting changed pixel data when the amount of change is small. .

[Disclosure of Invention]

Embodiment 1 Shows a circuit on the transmitting side and a circuit on the receiving side in the embodiment of the present invention shown in FIGS.
The sync signal is obtained from the sync signal generation circuit 44, and image transmission is controlled and controlled by the sequencer 45. When transmitting a new image, it is called a 64 × 64 pixel image → 128 × 128 pixel image → 256 × 256 pixel image. As described above, the image quality is gradually improved from a coarse image to a fine image, and the state of change can be transmitted as soon as possible. In this case, parameters such as the number of pixels to be transmitted, a compression parameter (necessary for selecting a compression mode described later), and full screen / changed part selection are obtained from the parameter table 46. The video signal from the TV camera 1 is converted into a digital value (usually 6 to 8 bits) by the A / D converter 3 and written in the frame memory 6 indicated by the address counter 4. In the embodiment, the standard is 256 × 256 pixels. The frame memory 6 is accessed by an arithmetic control circuit 47 such as a CPU, and the address selector 50 is for switching between the address data of the arithmetic control circuit 47 and the address data from the address counter 4. The current image sample output from the frame memory 6 is loaded into the current image frame buffer 37. The current image frame buffer 37 can be shared with the frame memory 6, but in the embodiment, it is separately provided in consideration of the case of using a plurality of TV cameras 1. The data of the current image written in the current image frame buffer 37 is the next change detection circuit.
At 48, a two-step test is performed to determine whether it has changed with respect to the transmitted previous image and whether it has changed with respect to the reference background image. The previous image data buffer 49 is a buffer for writing the data of the transmitted previous image (not the predicted image) used for the test, and the reference image data buffer 51.
Is a buffer for writing the background image data that does not change and serves as the above-mentioned reference. Of course, only the change detection for the previous image may be set, and in this case, the inter-frame residual is smaller than that for the reference background image.

FIG. 2 shows a flow chart of the operation relating to the change verification of the change detection circuit 73. As can be seen from this flow chart, sequence control is performed. The magnitude of the change for each pixel is sequentially obtained by the absolute value of the difference between the surface level image (for example, extracted every four pixels) and the corresponding pixel on the current screen. Then, it is examined whether or not the magnitude of the change for each pixel exceeds a predetermined first reference value R ₁ .
The cumulative addition value (projection [projection]) of the magnitude of the change to the X coordinate and the Y coordinate is obtained, and the cumulative addition value is larger than the second reference value R ₂ The coordinates (X ₁ , Y ₁ ) ( X ₂ , Y ₂ ) is calculated in the same way as in the case of FIG. It is possible to specify whether the coordinates of this change area are finally for the previous image or for the reference background image. If the change detection sequence is first for the reference background image, then for the previous image, then for the previous image. The coordinates are obtained. Now, the change area obtained next is the third predetermined area.
Is larger than the reference value R ₃ , and when larger than the reference value R ₃ , it is determined that there is a change, and the determination result and the coordinate data of the changed area are output. As a result, the output is fetched by the compression method switching circuit 52, and the compression method switching circuit 52 can switch the compression method according to the size of the changing area or the number of changing pixels, as described later.

It should be noted that, in order to obtain the conversion area separately from the reference value R ₃ indicating the presence or absence of change, it is possible to optimize each by dividing it into another reference value R ₃ ′ (R ₃ ′ = 1/2 · R ₃ ).

The changed pixel determination circuit 53 is a primary prediction unit based on the current image frame from the current image frame buffer 37 and the transmitted previous image.
When the inter-frame residual between the predicted image frame buffer 54 and the predicted image frame of the 55 is a predetermined fourth reference value ± R ₄ or more (absolute value of the difference is R ₄ or more), the pixel is changed and R ₄ If it is less than zero, it is treated as a zero pixel, and if one line is all zero pixels, it is treated as a zero line. FIG. 3 is an explanatory diagram of the above-described test, and FIG. 3A shows the inter-frame residual,
The same figure (b) has shown the residual between frames after a test. Here, the reference value R ₄ is set to be smaller than the reference value R ₁ used for the change detection, which allows the portion outside the rectangular area of the change area by the change detection to be transmitted as a change pixel, and this rectangle It is possible to avoid transmitting many zero pixels in the area. That is, the sensitivity of change detection can be lowered, for example, the change in the brightness of the room to be monitored and the influence of the automatic aperture of the TV camera 1 can be reduced, and the lack of extraction of changed pixels for image transmission can be eliminated. . Here, the reference value R ₁ is an optimum value with or without change,
Needless to say, the reference value R ₄ is set to an optimum value for determining the changed pixel.

Now, in the change pixel determination in the change pixel determination circuit 53, the change pixel / zero pixel is made to correspond to the logical value 1/0, respectively, and the isolated point is degenerated and removed by the presence / absence of image connection as a binary image, so that the isolated pixel in the change pixel is isolated. The dotted zero pixels propagate the changed pixels and remove them so that the changed area can be transmitted efficiently and correctly. Therefore, in the circuit, a binary image frame buffer and a logic for removing isolated points by connection detection are provided. Built-in arithmetic circuit. The compression method switching circuit 52 is for determining which percentage of the previous screen is in the change area and selectively switching the compression method.
And 20% are set, and if the number of changed pixels is less than 20% due to the width (width) of the rectangular area obtained by change detection, zero pixels, zero line compression and expansion are performed, and 20% or more. In this case, the variable sample density compression / expansion is used to perform predictive coding (logarithmic compression) and compression in the time direction according to the quantization characteristics shown in Tables 1 and 2. In the embodiment shown in FIG.
When applied to the range of 20% or more and up to 50%, the inter-frame residual is compressed in the one-dimensional direction, and when it exceeds 50%, the inter-frame residual of the entire screen is compressed in the two-dimensional direction. This primary
The secondary switching is performed when the secondary predictor is provided. In the embodiment of FIG. 1, the secondary predictor 56 is provided so that the secondary prediction can be performed. Here, the primary prediction unit 54 includes a frame pre-value prediction circuit 57 and the predicted image frame buffer 55, and a variable sample density compression circuit 28 'and a variable sample density decompression circuit 29' provide a primary inter-frame variable sample. The secondary prediction unit 56 is composed of a predictive value prediction circuit 58 including a prediction frame buffer and a prediction residual line buffer 59, and the intra-frame variable sample density prediction coding circuit. Combined to form a circuit equivalent to the secondary inter-frame variable sample density predictive decompression decoding circuit.

Thus, in the compression method switching circuit 52 described above, the number of changed pixels is
If it is determined to be less than 20%, the current image frame is compressed and transmitted by the coding compression method of zero pixels and zero lines based on Table 4, for example. That is, the data taken into the change pixel line buffer 60 is converted into an 8-bit code by the zero pixel / zero line compression circuit 61 when the upper 2 bits are “00” as shown in Table 4, and when the value is “01”, the change pixel is zero pixel. Number, the number of zero lines when "10", and the line end code or the frame end code when "11" are continued, and the lower 6 bits are the signed 6-bit data of the inter-frame residual of the change pixel. Alternatively, the upper 6-bit data of the changed pixel, the number of zero pixels, the number of zero lines, the line end code, or the frame end code is inserted. In this case, the change area is
It is possible to faithfully transmit the information of the changing pixel when the width is narrow at about 10%. At the same time, the decompression circuit 62 replaces the value of the non-zero pixel of the predicted image predicted by the previous frame value prediction circuit 56 with the value of the current image.

Next, when the number of changed pixels exceeds 20% Compression method switching circuit 52
Signal switching by the switching units 63, 64, 65, and when the range does not exceed 50%, the primary / secondary switching circuit 66 disconnects the secondary predicting unit 56, and if the range exceeds 50%. In this case, the secondary prediction unit 56 is connected. In each case, the variable sample density compression coding of the inter-frame residual is performed. In this case, since the correlation between pixels of the inter-frame residual is small, the quantization characteristic with a small time difference value as shown in Table 2 is used. That is, in the case of Table 1, the time difference value is 8 when the quantization level is “7”, and the time difference value is 4 when the quantization level is 6, 8; Although a position error is generated, in the case of Table 2, only a maximum of 1 pixel causes a position error.

Even if the time difference value when the amplitude difference value is 0 as shown in Table 2 is 2 as shown in Table 3 for example, the code compression circuit 68 has zero pixels,
Code compression is performed so that a code having an amplitude difference value of 0 on a zero line can be compressed to improve transmission efficiency. The number of pixels to be transmitted, the compression parameter and the total image / A code header for indicating the changed portion selection is added to each frame and transmitted as communication data via the packet data editing circuit 66. The code compression circuit 68 may determine the code amount, and if the code amount is large, the compression method may be switched again. FIG. 1 (b) shows a decoding circuit on the receiving side of the embodiment corresponding to the transmitting side of FIG. 1 (a). The transmitted communication data is transmitted by the packet data separation circuit 69 to the number of transmission pixels and compression parameters. , The whole image / change part selection code is separated and loaded into the parameter table 70,
Data indicating the transmission of the number of zero lines or the transmission of the residual between frames is input to the compression system switching circuit 71, and the switching units 72 and 73 are operated by the switching control from the compression system switching circuit 71 to output signals respectively. It is distributed to the corresponding expansion circuits 74 and 75. In addition, the first-order
The secondary switching circuit 85 switches between configuring a primary inter-frame variable sample density predictive compression / decompression decoding circuit or a secondary inter-frame variable sample density predictive compression / decompression decoding circuit,
The connection between the primary prediction unit 76 and the secondary prediction unit 77 is switched. The primary prediction unit 77 is composed of an inter-frame prediction circuit 78 and a prediction image frame buffer 79, and the secondary prediction unit 77 is composed of a prediction circuit 80 and a prediction residual line buffer 81. The decoded image data is displayed as an image on the display device 84 via the display image frame buffer 82, and stored and stored in the stored image memory 83.

Second Embodiment In the first embodiment, the number of changed pixels is divided into three stages, and primary and secondary predictive compression encoding is performed when transmitting the inter-frame residual. However, in the present embodiment, only the primary compression is performed. The compression encoding and decoding circuits are configured as shown in FIGS. 4 (a) and 4 (b). The same reference numerals as those in the circuit of FIG. 1 have the same configuration and operation.

In the case of full image transmission, the variable sample density compression circuit 28 ', the expansion circuit 29', and the secondary prediction unit 56 are passed, but the change pixel is not determined.

〔The invention's effect〕

The present invention uses the transmitted previous image frame as a reference image, obtains the difference between the image frame of the reference screen and the current image frame, and compresses only the image of the changed area for narrow band transmission. Therefore, the receiving side has only the previous image and there is no fixed background image.Because it compares the previous image frame and the current image frame that have already been transmitted, there is little fluctuation in the amount of transmission each time, and the frame update cycle is It can be kept short below a certain level, and when the density information of each pixel is transmitted as N-bit (N is an integer) pixel data, the magnitude of the difference between each pixel and the reference screen is obtained. And a step of testing the magnitude of change for each pixel where the magnitude of the difference is zero when the magnitude of the obtained difference is less than or equal to a first reference value that can be regarded as significant for change detection, Pixel A step of adding the size of each pixel size of the change from the results of the assay for each line X, in each direction of the Y accumulated among the accumulated value exceeds the second reference value, X, Y
The process of detecting the respective minimum and maximum values in the direction of, and the size of the rectangular area determined by the respective minimum and maximum values,
In the process of determining that there is a change when the third reference value is exceeded and the process of transmitting the image information, the inter-frame residual error, which is an error between the current image frame and the predicted value based on the transmitted previous image frame, is described above. The process comprises transmitting a pixel value of a current image frame that is smaller than a first reference value and exceeds a fourth reference value, and detects an interframe residual error of a pixel whose interframe residual is smaller than the fourth reference value. Since it is transmitted by variable sampling density encoding within the frame as zero, change detection and change area detection are performed accurately, and the sensitivity of change detection is reduced, and changes in room brightness and automatic aperture of TV cameras are performed. The effect of can be reduced, and the removal of the isolated points of the changed pixels can eliminate the omission of the extracted changed pixels for image transmission, thus improving the image quality in the changed basin.

[Brief description of drawings]

FIG. 1 (a) is a circuit configuration diagram on the transmitting side according to the first embodiment of the present invention, FIG. 1 (b) is a circuit configuration diagram on the receiving side, and FIG. FIG. 4 is a time chart for explaining the operation of the above, FIG. 4 (a) is a circuit diagram of a main circuit on the transmitting side according to the second embodiment of the present invention, and FIG. 4 (b) is a circuit diagram of the main circuit on the receiving side. 5 and 5 are circuit configuration diagrams of a conventional example, FIG. 6 is an operation explanatory diagram of the same as above, FIGS. 7 (a) and 7 (b) are main circuit configuration diagrams of the same as above, and FIGS. Is an explanatory diagram of the variable sample density coding system, FIG. 9 is an explanatory diagram of the variable sample density predictive coding system, and FIGS. 10A and 10B are interframe variable sample density predictive compression encoding circuits and interframe variable FIG. 10C is a circuit configuration diagram of the sample density prediction decompression decoding circuit, FIG. 10C is a circuit configuration diagram of another inter-frame variable sample density prediction compression encoding circuit, and 48 is a change. Knowledge circuit, 52 a compression system switching circuit 53 is changed pixel determination circuit, the primary prediction unit 54, 56 the secondary prediction unit, R ₁ to R ₄ is a reference value.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-51-80711 (JP, A) JP-A-59-152785 (JP, A) Takahiko Fukibuki "Digital Signal Communication for Images" (Sho-58) Newspaper company) P. 208-216

Claims (3)

[Claims] 1. A method of performing narrow band transmission by compressing information of only an image of a changed area by obtaining a difference between an image frame of a reference screen and a current image frame by using a transmitted previous image frame as a reference image. In the process of transmitting the density information of each pixel as N-bit (N is an integer) pixel data, the step of obtaining the difference between each pixel and the reference screen,
The magnitude of the obtained difference can be regarded as significant for change detection.
When the difference is zero when the difference is less than the reference value, the process of testing the magnitude of change for each pixel and the result of testing the magnitude of change for each pixel The process of cumulatively adding the magnitudes in each of the X and Y directions, and the process of detecting the respective minimum and maximum values in the X and Y directions among the cumulatively added values exceeding the second reference value, The process of determining that there is a change when the size of the rectangular area determined by the minimum value and the maximum value exceeds the third reference value and the process of transmitting the image information are performed on the current image frame and the transmitted previous image frame. And a value of a pixel of the current image frame in which the inter-frame residual error, which is an error from the predicted value based on the fourth reference value, is smaller than the first reference value is transmitted. Within the frame, the inter-frame residual of pixels smaller than the fourth reference value is set to zero. Image transmission method characterized by transmitting the variable sampling density coding. 2. An interframe residual whose interframe residual is larger than the fourth reference value or a pixel value of a current image frame is transmitted as it is, and is an interframe residual smaller than the fourth reference value. When the value of the image frame is zero, the number of pixels that become zero and the number of lines in which all the pixels become zero are encoded and transmitted, and the values of non-zero pixels of the predicted image frame are The image transmission method according to claim 1, wherein the image value is replaced with a pixel value. 3. A variable sample density predictive coding within a frame is transmitted by setting the inter-frame residual of pixels whose inter-frame residual is smaller than the fourth reference value to zero. The image transmission method described in the item.