FI92271C - encoding method - Google Patents

Description

92271

encoding method

The invention relates to an encoding method as defined in the preamble of claim 1, in particular for encoding a live video signal for a narrow transmission channel.

The transmission of a video signal typically requires a dialing channel with a blade capacity of 2 to 70 Mbit / s.

When the video signal is compressed by detecting changes or differences in the picture fields of perSk-10 by encoding and transmitting them, the transmission channel capacity can be dropped to between 64 Kbit / s and 2 Mbit / s.

Known methods for compressing a live video signal are generally based on the procedure of dividing an image field 15, which consists of e.g. 256 x 256 pixels, i.e. a list of pixels, into groups of pixels, i.e. clusters, e.g. containing 8x8 pixels. By using clusters, the number of address bits to be transmitted can be reduced. The address of a single pixel, typically defined by 6 bits, located at an arbitrary position in the 256 x 256 pixel image field, requires 16 bits. For clusters, such as 8x8 pixel clusters, a 10-bit common address is used for all pixels in the cluster, such as 64 pixels.

25 The disadvantage of cluster-based coding methods is their ability to handle the edge areas of moving objects efficiently and accurately.

They either waste the capacity of the transmission channel by retransmitting the permanent information or they cause the edge of the moving object to become obstructed. dirty window effect.

The object of the invention is to provide a new coding method by means of which the disadvantages of the above-mentioned Klus-based coding methods can be eliminated. The object of the invention is in particular to provide a new coding method by means of which perSk-band picture fields are coded on the basis of pixels and in which the address of the pixels is compressed in such a way that a clear contamination in the channel capacity is achieved.

The method according to the invention is characterized by what is stated in claim 1.

In the encoding method according to the invention, in particular for encoding a live video signal across a narrow transmission channel, each image of the video signal to be transmitted is sequentially processed by retrieving a previous reference image stored in the image memory and comparing the images with each other. According to the invention, each image is kSsi-15 in pixel format; the image and the reference image are compared pixel by pixel and a difference image is formed based on the differences in the gray tones of the pixels; the difference image consisting of a continuous grayscale scale is quantized according to a preset grayscale scale into a plurality of 20 sub-images, each sub-image being a binary image, each pixel having a binary value of 1 or 0, depending on whether a certain amount of grayscale is not valid; all subpictures are then encoded using commonly known-25 and binary image encoding methods and transmitted via a transmission channel to the receiver; The binSS images are received at the receiver and decoded by a corresponding binary image decoding method, after which the image information 30 obtained from the image memory of the receiver is updated with the received difference information pixel by pixel and read from the image memory as a video signal for further processing.

In one application of the method, the difference image is divided according to the size of the grayscale difference into 35 limited numbers 2k + 1 independent subpictures, where • Λ: 92271 3 • difference 0 zero 5 1 smallest difference (positive) 2 second smallest difference (positive) 3 third smallest difference (positive)

• · I

k maximum difference (positive) 10 k + 1 smallest difference (negative k + 2 second smallest difference (negative) • »· 2k maximum difference (negative) 15 and where each sub-picture 2k + 1 comprises nxm pixels, each pixel having binary value 1 or 0 depending on whether a certain amount of grayscale difference is valid in the pixel in question or not.

In one embodiment of the method, the sub-images 20 are encoded with a pass code. In the run code, bits 1 and 0 are grouped into suitable groups having the same binary value. The number of ones or zeros in the groups is reported as a number through the forward transmission channel to the recipients.

25 In one application of the method, a bridging method is used in the running code to shorten the coding. This means that separate inverse binary values in a given group of ones or zeros are eliminated and interpreted as the same binary values as the other members of the majority of the group.

In one application of the method, the color image is known to be divided into three sub-color images, such as RGB or YUV, each of which is treated separately as its own grayscale image. In this way, the color video image can be transmitted to the transmission channel encoded by the method according to the invention.

Regarding the advantages of the invention, we state the following. Although the 92271 4 image is quantized into a set of binary images, the coding of the address information is significantly saved so that the amount of information input to the transmission channel is clearly reduced. A further advantage of the invention is that the quantization of the image into binary images is a straightforward operation and no distortion can occur. All grayscale differences are transmitted according to the quantization steps, and a similar exact copy of the difference image is produced at the receiver. A further advantage of the invention 10 is that if limited distortion is accepted, the bridging method can be used efficiently in connection with the compression method according to the invention.

The invention will now be described with reference to the accompanying drawing, in which Figure 1 schematically shows the conversion of a grayscale image into a set of binary images; Figure 2 shows a block diagram of a transmission using the method according to the invention; and Figure 3 shows 20 receivers using the method according to the invention.

The digital video signal containing the image information to be transmitted by the pulse code modulation method requires a transmission channel capacity, the size of which is determined by the number of pixels to be transmitted, grayscale and the number of images to be transmitted in the municipalities. The information to be transmitted can be reduced without degrading the image quality by transmitting only a video signal representing a change between successive images. In this way, one way or another, one has to look in the image for the areas that have changed compared to the previous 30 images, which are thus sent to Aino-ast. In addition, the information of the transmitted changed image areas must be provided with addresses indicating which area of the image is in question. This makes proper image reconstruction possible. Eras 35 Such a video compression method is disclosed in Finnish patent publication 70662.

A pixel, or pixel, located in an imaginary position in the image field, can be represented by the following data: pixel value (gray level) + pixel location address (location coordinates, i.e. x-address + y-address). For example, if the image is 256 x 256 pixe-5, the total pixel address is 16 bits (both the x and y addresses require an 8-bit address) and the pixel value is mapped at 64 grayscale levels (6 bits), then the amount of information to be transferred by tSlld is next: 6 bits (pixel value) + 16 bitsS (address) = 24 bitsS / pixel.

10 From this example, it can be seen that a large part of the transmission capacity is dedicated to the transmission of pixel addresses (75% in the example).

The coding method according to the invention is directed in particular to the addresses of the pixels. According to the invention, in the 15-pixel coding method, it is precisely the compression of the addresses that is carried out in such a way that considerably fewer bits per pixel are sent to the transmission path. The invention relates to an image conversion method in which a grayscale image is converted into an organo-20 set of binary images which carries address information and which can be further processed using binary image coding methods known per se.

The method according to the invention is described below with reference to Figure 1. The method according to the invention utilizes a video compression method in which only a video signal representing the change between the percussion images is transmitted to the transmission channel. Thus, from the grayscale image 1 and the reference image 2 of the video signal 2, a difference image 3 is formed. All said images are formed of n x m pixels, where n 'and m are integers. The difference image 3 is formed by comparing the grayscale image of the video signal from 1 pixel pixel to the corresponding pixels of the reference image 2. Only the quantized image 4 is formed in the difference 3. TMmå means that the difference image 3 is divided according to a predetermined gray scale scale into a limited number of margins 2k 92271 6 + 1, where k = an integer, independent sub-images.

This can be presented in the form of the following table: # difference 5 0 zero 1 smallest difference (positive) 2 second smallest difference (positive) 3 third smallest difference (positive) 10 k largest difference (positive) k + 1 smallest difference (negative) k + 2 second smallest difference (negative) • a «15 2k largest difference (negative)

Each subimage 2k + 1 comprises n x m pixels, each pixel having a binary value of 1 or 0. The binary value of 1 describes the existence of 20 states of a certain amount of grayscale separation. A binary value of 0 describes the absence of this difference.

The grayscale of the image is then, for example, divided into 64 levels, and during the quantization process the grayscale of each pixel is checked and then classified into a specific sub-image corresponding to this grayscale. The other sub-25 differences have a binary value of 0 for this pixel. Thus, all sub-images have n x m pixels with a binary value of 0 or 1 for each pixel.

All binary images 2k + 1 realized as described above can now be coded using coding methods known per se, which are specifically suitable for coding binary images. We refer to e.g. publications: 1) Hunter, R. & Robinson, A.H .: International Digital Facsimile Coding Standards, Proc. Of IEEE, Vol.

35 7, July 1980, p. 854-867; and 2) Yasuda, Y .: Overview of Digital Facsimile Coding Techniques in Japan, Proc. of IEEE, Vol. 7, J »92271 7

July 1980, ss. 830-845.

Each binary image is encoded in the preferred code kSyt, thus benefiting the run code. TarnM means that the binary values of each binary image are grouped into groups of zeros 5 and ones, from which the binary value and their number are sent to the successive transmission channel. Such running codes are in the general kSytds e.g. fax lalttelssa.

In connection with the coding method according to the invention, a bridging method can be used, and limited distortion in image transmission is allowed. By bridging method is meant a general process by which the coding of a pass code can be improved. In the bridging method, e.g., bit "0" is changed to bit "1", 15 if the two bits preceding it are both "1" and the next two bits are also "1". For example, if the bit string is: 11011, the selectable bit "0" is changed to one when encoding the run code. It should be noted that bridging yields can vary considerably.

The coded binary images are received by a receiver, where the used running code is decoded and the information corresponding to the difference image is obtained. The difference image is compared with the reference image of the receiver's image memory pixel by pixel, and a new image is formed, which is stored in the image memory. This image is still available in the receiver's memory.

A video signal transmitter for carrying out the method according to the invention is shown in Figure 2. The video signal is input to a receiver via an A / D converter 6, in which converter it is converted to digital form. The digital video signal is further input to the first picture memory 7. A reference picture is stored in the second picture memory 8. In each image memory 7, 8, the image is in pixel format as an n x m matrix. The digital video image 35 is input from the first memory 7 pixels at a time to the comparator 9 simultaneously and at the same stage as the second image memory 8 is also fed to the comparator 9. The comparator 9 generates a difference image which is further fed to the includes the grayscale scale of Table 5 stored in ROM. This table includes 2k + 1 shades of gray lime according to the table above. The difference image input to the grayscale comparator 10 is compared with the grayscale stored pixel by pixel, and the image is divided on this basis into sub-images, i.e., 10 binary images, a total of 2k + 1 of lime, or as many as the number of grays stored in the ROM. Each sub-picture 2k + 1 has its own memory unit 11; li1 - ll2k + 1. These memory units are preferably RAMs. The memory units 11 of the sub-images 15 are each connected to their own flow code 12; 12 to 12 and further to the selector 13 connected to the transmission channel 14. The subpicture memories 11 are run one by one in a sequential order and the bitmaps of the binary images containing 20 n x m bits to the selector 14 are further encoded by the sequence encoders 12 and coded 12.

A receiver applying the method according to the invention is shown in Fig. 3. The coded video signal from the transmission scheme 14 is input to the decoder 15 and then to the comparator 16. The comparator 16 is input with the received decoded video signal. keskenåån. The resulting new image data input-30 is then returned to the image memory 17 and thus the image memory is updated. The updated image memory 17 is decompressed through the D / A converter 18 and a video signal is thus generated, which is fed forward to a display device or a suitable image processing device or the like.

35 The above-mentioned transmitters and receivers further comprise a control unit 19 and 20, respectively, by means of which the transmitter and receiver, respectively,

-> »K

92271 9 is controlled appropriately.

The invention has been described above with reference to only one preferred embodiment thereof, but it is clear that the invention can be modified in many ways within the scope of the appended claims.

Claims (5)

92271 10 A coding method for encoding a live video signal in particular over a narrow transmission channel, wherein each image of the video signal to be transmitted is sequentially processed by retrieving a previous reference image stored in the image memory from the memory and comparing the images, and transmitting only the changed image areas. to a receiver-10, wherein - each image is processed in pixel format (nxm); and - comparing the image and the reference image with each other pixel by pixel and generating a quantized difference image based on the differences in their gray tones; characterized in that - a difference image consisting of a limited quantized grayscale scale is converted into a plurality of sub-images, each sub-image being a binary image in which each pixel has a binary value of 1 or 0 depending on whether a difference value of 20 is valid at that pixel; - all the sub-pictures are encoded using the binary picture encoding method and sent via the MS transmission channel to the receiver; - the binary images are received at the receiver and de-25 are encoded by a corresponding binary image decoding method, after which the image information obtained from the image memory of the receiver is updated with the received difference information pixel pixelH and read from the image memory as a video signal for further processing. A method according to claim 1, characterized in that the difference image is divided according to the magnitude of the gray scale difference into a limited number of 2k + 1 independent sub-images, wherein: 35 92271 11 # difference 0 zero 1 smallest difference (positive) 5. second smallest difference (positive) 3 third smallest difference (positive) k largest difference (positive) k + 1 smallest difference (negative) 10 k + 2 second smallest difference (negative) • · · 2k maximum difference (negative) and where each subimage 2k + 1 comprises nxm pixe -15 liia, each pixel having a binary value of 1, which describes the validity of a certain amount of grayscale separation in question. in pixels, or 0, which describes the absence of this difference. Method according to Claim 1 or 2, characterized in that the subpictures are coded with a running code (RLC). A method according to claim 3, characterized in that a bridge method is used in the running code to shorten the coding. Method according to one of the preceding claims, characterized in that the shadow image is divided into three sub-color images (RGB or YUV), which are known per se, each of which is treated separately as its own grayscale image. 30 92.71 12