JP3303869B2 - Image encoding method, image encoding device, image decoding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to communication of television signals (hereinafter referred to as TV signals) and image signals.
The present invention relates to a conference device and a TV telephone device.

[0002]

2. Description of the Related Art Digitally compressing a TV signal to 64k
2. Description of the Related Art TV telephones and TV conferences that use a transmission path of about b / s for transmission are becoming widespread. However, mutual communication was impossible because the format of the screen, the compression method, and the like differed from one manufacturer to another. Therefore, the International Telegraph and Telephone Consultative Committee (CCITT) studied standardization of compression methods and screen formats, and came up with recommendations.

The outline of the CCITT standard system will be described below.

In the standardized system, as shown in FIG.
Intermediate format (CIF: Common Intermediate Forma
An image format called t) is used. CI
F indicates one frame of the TV screen as having a luminance signal of 352 × 288 pixels, a color signal of CB176 × 144 pixels, and a color signal of CR17.
It is digitally divided into 6 × 144 pixels,
The transmission is performed at 0 frames / sec. The CIF is an intermediate between the NTSC image format, which is a TV system mainly in Japan and the United States, and the PAL format, which is mainly in Europe.

[0005] As shown in FIG.
It is divided into regions called GOBs (Group Of Blocks). The GOB is further divided into 33 macro blocks (MB). That is, the MB is composed of a luminance signal of 16 × 16 pixels, a color signal CB of 8 × 8 pixels, and a color signal CR of 8 × 8 pixels. The luminance signal is further divided into four 8 × 8 pixel blocks, and together with the two color signals, becomes a six 8 × 8 pixel block.

The compression encoding process is performed by the source encoding circuit 6 and the video multiplexing circuit 7 shown in FIG. The source encoding circuit 6 suppresses the redundancy of the image, and the video multiplex encoding circuit 7 encodes and transmits the pixel signal from which the redundancy has been removed.

[0007] For details of the source code and the video multiplex coding, see "PCSJ89 Image Coding Lecture" hosted by the Institute of Electronics, Information and Communication Engineers, Japan.
It is described in detail from page 5.

In source coding, a screen (reference screen) coded and transmitted immediately before is reproduced and held by itself. For each MB to be coded, the most similar part (the size is the same as the MB, hereinafter referred to as MB) is searched from the reference screen, and the difference between the MB to be coded and the reference MB is transmitted. The difference indicates a signal obtained by subtracting the value of the corresponding pixel in the reference MB from the value of each pixel in the encoded MB. This difference signal is subjected to orthogonal transform called discrete cosine transform (DCT) for each 8 × 8 pixel.
Miyahara, "Systematic Image Coding" (IPC, 199
0.7) from page 222, and 250
As described from the page, DCT is a type of peripheral number transformation, and distributes an image signal or its difference signal into frequencies (DCT coefficients). The compression is performed by transmitting the low frequency component of the DCT coefficient obtained as described above.

At the same time as transmitting the DCT coefficient,
Is also transmitted as a “motion vector”. If the values of the corresponding pixels of the MB to be coded and the reference MB are almost equal, the DCT coefficients are all zero and are not transmitted (NOCODED). Of the NOCODED, when the motion vector is particularly zero, that is, when there is no change in the image of that part, only the information of the MB without change (invalid MB) is transmitted.

In video multiplex coding, a signal resulting from DCT, MB attributes (such as NOCODED, etc.), motion vectors, and the like are coded using variable-length codes and transmitted.

The above is the outline of the standardization method.

[0012]

Generally, when an NTSC signal is digitized, it is an integral multiple of about 3.58 MHz, which is a reference signal inside the NTSC (generally, about 14.3 MHz which is four times as large as the reference signal).
Since sampling is performed at the frequency of z), 910 pixels are provided per one scanning line. The number of scanning lines (resolution in the vertical direction) of one frame is 525 lines. TV phone or TV
In a conference or the like, only the effective area of these signals excluding the vertical / horizontal synchronization signals is transmitted. That is, horizontal 76
A format of 8 pixels × 480 lines or 384 pixels × 240 lines of half horizontal and vertical resolution is used. In particular, when a very slow transmission line of about 64 kb / s to 384 kb / s is used, the latter 38
A 4 × 240 format is often used. For this reason, when communication is performed in a standardized system, the NTSC TV
The signal must be converted to CIF and compressed / encoded. Conversely, the receiving side also reproduces the reproduced CIF signal as NT
It must be converted to an SC signal and displayed.

FIG. 11 shows an example of a typical encoding apparatus.
First, the input TV signal (NTSC)
In the sampling circuit 2, the number of pixels in the horizontal direction is CIF (35
(2 pixels). A number of methods are conceivable for this conversion, but typical ones are shown in FIGS. FIG. 7 shows that the analog TV signal is separated by the color / luminance separation circuit 20 and then sampled by the A / D converters 21 and 24 at a frequency (luminance: about 6.7 MHz, color: about 3.4 MHz) suitable for CIF. Method, FIG. 8 shows A / D
The TV signal digitized by the converter 21 is separated into a color signal and a luminance signal by a digital color / luminance separation circuit 23,
This is a method of converting the sampling number by the digital filters 22 and 25.

The signals sampled in this manner are sequentially read out in the vertical direction in the order of a luminance signal, a color signal CB, and a color signal CR.
A new pixel is added at the ratio of.

The scanning line number conversion filter produces six pixels from five pixels read in the vertical direction. FIG.
FIG. 1 shows the principle of scanning line number conversion by linear interpolation. In FIG. 13, for the j line and the (j + 6) line of the CIF, the values of the i line and the (i + 5) line of the NTSC are output as they are. The (j + 1) to (j + 5) lines are each calculated by a weighted average of values of upper and lower lines in space. For example, the (j + 1) line is NT
It is located at the 5/6 position between the SC i line and the (i + 1) line. Therefore, 1/6 times the signal of (i line) and (i
+1) by adding 5/6 times the signal of the line
The signal of the IF (j + 1) line is obtained. FIG. 12 shows an example of the scanning line number conversion filter.

240 lines TV read out vertically
The signal 102 is together with the previous signal 121 held in the delay circuit 70. Used to calculate the weighted average described above. The signals 102 and 121 are weighted by the weighting circuit 71 and the weighting coefficients 124 and 125 generated corresponding to the read vertical address 114 of the image signal and the integrators 72 and 125, respectively.
73, a weighted average is calculated in an adder 74, and is converted into a 288 line TV signal 104. This processing is input 2
The 40-line signal is completed in a 5-line cycle, and the data held in the delay circuit 70 is also synchronized with the signal line 12 in synchronization with this cycle.
Cleared by 6.

As another method of converting the number of scanning lines, a method of applying an interpolation filter in the vertical direction or a screen having 480 lines of scanning lines is generated by sequential scanning as seen in a current clear vision receiver. 48
There is a method of creating 288 scanning lines by thinning 5 lines to 3 lines from 0 lines.

On the receiving side, the scanning lines transmitted by 288 lines are thinned out by 6 to 5 to output 240 lines.

As described above, the number of scanning lines must be converted even in the communication between the NTSC systems, which causes problems such as deterioration of the image quality due to the conversion, and the circuit scale and cost of the conversion device itself.

Accordingly, it is an object of the present invention to encode a TV signal by a method conforming to standardization without deteriorating image quality due to conversion of the number of scanning lines.

[0021]

An object of the present invention is to transmit a valid TV signal only to 240 lines of a screen (288 lines) transmitted by CIF and pack invalid information into the remaining portion (48 lines). This is achieved by:

According to the above-described method, the deterioration of the image quality due to the conversion of the number of scanning lines during communication between NTSCs is eliminated, and the circuit scale is reduced as compared with the case where a scanning line number conversion circuit is provided. Furthermore, since the number of pixels to be encoded and transmitted is reduced, the amount of information per pixel can be increased, and a TV signal with better image quality can be transmitted. In addition, since the processing time per pixel also takes a long time, the device is slower than the conventional device,
Alternatively, it is possible to adopt a simpler circuit configuration than before, or to encode and decode a larger number of frames per unit time than before using the same elements and the same circuit configuration as before.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail using a screen.

FIG. 1 is a block diagram of a TV signal encoding apparatus incorporating the present invention. Part 10 is the part of the present invention. The input NTSC signal 1 is converted to a digital TV signal 101 in a color / luminance separation and sampling circuit 2. The digital TV signal is converted to a signal rearrangement circuit 3
, Only the portion of the effective area (240 lines) is cut out. As shown in FIG. 9, the signal rearrangement circuit 3 has a built-in memory 30 capable of storing one frame.
The time axis is converted in the order of the color signal CR and converted into a 240-line digital TV signal 102. On the other hand, the fixed value sending circuit 4 outputs an arbitrary digital fixed value to the fixed value signal 103. As shown in FIG. 10, the fixed value sending circuit 4 includes a ROM 40 storing the values of the pixels in the fixed area so that an arbitrary value can be repeatedly output. This data is specified by the address signal 112. This fixed value may be any value as long as it is within the dynamic range of the image. The 240-line digital TV signal 102 and the fixed value signal 103 are switched and synthesized by the switch 5, and
8 lines. This switching signal 110 is generated by the timing circuit 9. 28 generated by switch 5
The 8-line digital TV signal 104 is compressed by the source encoding circuit 6 and the multiplex encoding circuit 7 and output to the transmission path 8.

The timing circuit 9 simultaneously outputs the address 11
1 and 112 are sent to the signal rearranging circuit 3 and the fixed value sending circuit 4, respectively. FIG. 4 shows a timing chart of the timing circuit 9. The upper 240 lines of the luminance signal select the signal of the 240-line digital TV signal 102 (the switching signal 110 is at the high level in the timing chart), and the remaining 48 lines of the luminance signal have the fixed value signal 1.
03 is selected (low level in the timing chart). Regarding the color signal, the upper part 120 corresponds to the luminance signal.
The line is set to high level, and the remaining 24 lines are set to low level. Thus, the CIF TV signal 104 is obtained. The encoding process is performed by a method completely conforming to the standardization as in the conventional example. The image to be coded is the upper 240 as shown in FIG.
A normal image signal enters the line, and an image determined by the fixed value signal 103 enters the lower 48 lines. When the luminance signal is fixed at 128 (dynamic range 0 to 255) and the color signal is fixed at 0, for example, the image of the fixed portion has a constant gray value.

At this time, reference numeral 106 transmitted to the transmission line 8
Is the GOB (1) corresponding to the upper 240 lines as shown in FIG.
To 10) contain the sign of the image, GOB11 and GOB
Reference numeral 12 denotes a code obtained by encoding the fixed value signal 103. Generally, the code amount of the fixed area is much smaller than the code amount of the image. For example, if the fixed area is encoded and transmitted once, and the source encoding circuit is in a mode for transmitting the difference from the previous image (normal encoding mode), there is no change from the image, so the code is not changed. The quantity will be 0 bits. As a result, the amount of information to be transmitted is reduced by 1/6 (about 17%) as compared with the case where all 288 lines are valid images.

On the receiving side, the code transmitted according to the standardization method is decoded to obtain a reproduced image of 288 lines. This is converted to 240 lines by scanning line number conversion,
When displayed, an image with a gray band below is obtained as shown in FIG. 5 (288 line display mode). If it is known that the other party has only the effective area of 240 lines, only the effective area may be displayed (240 line display mode). Means for informing the other party of the size of the effective area include inserting a special code after a specific (for example, eleventh) GOB header, notifying in a picture header, or a maker of an encoding device that transmits at the start of communication. It is conceivable that the judgment is automatically made based on the name or the like. In addition, when the lower two GOBs do not change continuously, they are automatically switched, or a switch for switching is attached to the device,
There are ways to let the user make a choice.

It is apparent that the following modifications are also included in the present invention. In addition, combinations of the following modifications are also included in the present invention.

In the above embodiment, the effective area of the screen to be transmitted has a size of 352 × 240, but the size of this screen can be any size as long as it is 352 pixels or less in the horizontal direction and 288 lines or less in the vertical direction. I do not care.

In the above embodiment, the DC value is used as the fixed area. However, if the image does not change between frames, communication can be performed with the same efficiency as in the above embodiment. In addition, even if there is a change in a part of the fixed area, if the change area is small, communication can be performed without much lowering the efficiency.

In the above embodiment, the fixed area is arranged at the lower part of the screen. However, the fixed area may be arranged at the upper part of the screen, or separated into the upper part and the lower part of the screen, or the effective area (24) of the screen to be transmitted.
(0-line TV signal) may be separated into a plurality of parts, and a fixed area may be arranged between the separated areas.

If it is determined that GOB 11 and GOB 12 are not transmitted for both transmission and reception before or during communication, the GOB headers of GOBs 11 and 12 need not be transmitted. In this case, the transmission efficiency is further improved.

In the 288 line image output mode, a fixed area is displayed on the screen. This fixed area portion may be replaced with an image generated on the receiving side and displayed.

The fixed area is (48) on the screen in units of 48 scanning lines.
(× N + 1) -th line (N is an integer), that is, in the case where they are arranged so as not to cross the boundary of GOB, the transmission efficiency is highest. However, even if the arrangement is performed on the (16 × N + 1) -th line (N is an integer) in units of 16 scanning lines, a high transmission efficiency can be obtained although the efficiency is slightly lower than the above-described arrangement. Further, it may be arranged on the (8 × N + 1) th line in units of 8 scanning lines.

In the case of "8 scanning line units", FIG.
As shown in FIG. 4, the boundary between the effective area and the fixed area of the screen is over the MB. In this case, the following processing is required.

The motion vector is fixed to zero in an MB having a boundary. Alternatively, when searching for a motion vector, a similar reference MB is searched using only the effective area portion (upper half: 16 × 8 pixels). When the difference signal searched and generated using only the effective area portion is encoded as it is, an extra image is added to a portion (lower half: 16 × 8 pixels) corresponding to the fixed area as shown in FIG. Enters,
Many codes are generated. Therefore, the portion corresponding to the fixed area is forcibly applied as shown in FIG.
NOCODED is used to prevent an increase in the amount of information, and the receiving side needs to perform processing to prevent deterioration of image quality by replacing this part with a fixed value.

In the color signal of the MB having a boundary, the boundary between the effective area and the fixed area also covers the block (FIG. 15 difference color block). If such a block is directly coded, many codes will be generated.
Therefore, the signal value in the fixed area is forcibly set to zero (see FIG. 1).
(5 correction color block left), or copy the effective area line-symmetrically to the fixed area centering on the boundary (FIG. 15 corrected color block right), or reconstruct the fixed area signal from the effective area by extrapolation prediction, etc. It is necessary to perform the encoding process after performing the above process.

[0038]

By applying the present invention, the picture quality of a TV signal to be transmitted can be improved and the circuit scale can be reduced, or more frames can be encoded per unit time. The effect of the implementation is extremely large.

[Brief description of the drawings]

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

FIG. 2 is an explanatory diagram of a CIF.

FIG. 3 is an explanatory diagram of GOB and MB.

FIG. 4 is a timing chart for switching image signals in the embodiment.

FIG. 5 is an example of an output image of the embodiment.

FIG. 6 is a code example of the embodiment.

FIG. 7 is a detailed diagram of a color / luminance separation / sampling circuit (Example 1).

FIG. 8 is a detailed diagram of a color / luminance separation / sampling circuit (Example 2).

FIG. 9 is a detailed diagram of a signal rearrangement circuit.

FIG. 10 is a detailed diagram of a fixed value sending circuit.

FIG. 11 shows an example of a conventional encoding circuit.

FIG. 12 is a detailed diagram of a scanning line number conversion filter circuit.

FIG. 13 is a principle diagram of scanning line number conversion.

FIG. 14 is an example in which an MB is applied to a boundary between an effective area and a fixed area.

FIG. 15 is a processing example when an MB is applied to an area boundary.

Continuation of the front page (56) References JP-A-1-258581 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04N ^7/ 24-7/68 H04N ^7/ 14-7 / 15 JICST file (JOIS)

Claims (11)

(57) [Claims] 1. An image encoding method for dividing an image into n parts for every a line and encoding the image, wherein a × n = N lines from an input original image signal having a number M of lines less than N lines. Is generated, and the generated signal of the N lines and the signal indicating the number M of lines of the original image signal are encoded. The encoded signal indicating the number M of lines is included in the header of the code of the image and transmitted. An image encoding method characterized by outputting to a road. 2. The image encoding method according to claim 1, wherein the signal of N lines is generated by adding a dummy signal of (N−M) lines to the input original image signal of the number M of lines. Image encoding method. 3. The image coding method according to claim 2, wherein said dummy signal is a signal of a fixed value. 4. An input original image signal is converted into an image signal of a size defined by a standard, the image signal is encoded, and information indicating an effective area occupied by the original image signal in the image signal is encoded. An image encoding method characterized in that the image information is output to a transmission path while being included in the header information of the converted image signal. 5. An original image signal having a number M of lines less than the number N of lines defined by a standard is input, and a fixed value signal for (N−M) lines is added to the original image signal to generate N lines. An image encoding method, comprising: generating a signal; encoding the generated N-line signal; including information indicating the number M of lines of the original image in a header of the encoded signal; 6. An image encoding apparatus which divides an image into n parts for every a line and encodes and outputs the divided parts, wherein an original image signal having a line number M less than the line number N of a × n = N is digitally converted. A sampling circuit that converts and outputs the image signal, a sending circuit that generates and outputs a dummy signal, and selects the digital image signal for the digital image signal and the dummy signal from line 1 to line M, For the following (N−M) lines, a switch that switches to select a dummy signal and outputs an N-line image signal, and a signal indicating the N-line image signal and a signal indicating the number M of lines of the original image signal are encoded. And a coding circuit that outputs the coded image signal. The coding circuit inserts a coded signal indicating the number M of lines into a header of the coded image signal. 7. The image encoding apparatus according to claim 6, the image coding apparatus, wherein said dummy signal is a fixed value signal. 8. The image encoding apparatus according to claim 6, wherein said encoding circuit divides said N-line image signal into macroblocks, and for each macroblock, from a reference image. The most similar part is searched, and the difference between them is DCT-transformed into D
A source encoding circuit for outputting a CT coefficient and a motion vector indicating a relative position thereof; a signal indicating the number of lines M of the original image is encoded and inserted into the header; And a video multiplex coding circuit for multiplexing and outputting the output. 9. An original image signal from line 1 to line M,
Subsequent (a × n−M) lines are a composed of dummy signals.
× n = receiving an image code obtained by encoding an image signal of N lines and information indicating the number M of lines of the original image signal contained in the header of the image code from a transmission path, decoding the image code, and reproducing the image code And displaying the area of the number M of lines from the top of the reproduced image. 10. The image decoding method according to claim 9, wherein said dummy signal is a fixed value signal. 11. An image code which is obtained by converting an original image signal into an image signal having a size defined by a standard and encoding the same, and transmitting a signal indicating the size of the original image signal included in a header of the image code. An image decoding method characterized by displaying an area of the size of the original image based on a signal indicating the size of the original image from above a reproduced image received from a road and decoding the image code.