JPH02302192A - image transmission device - Google Patents

Abstract

PURPOSE:To suppress the degradation of picture quality due to refresh by changing intervals of refresh in accordance with the performance of a transmission destination device. CONSTITUTION:When it is judged that the transmission destination has a good performance and has less error, intervals of refresh are extended. When it is judged that the error is large, intervals of refresh are shortened. Then, the reduction of the encoding efficiency is prevented. For example, error information 211 of an inverse DCT discrete cosine transforming circuit on the reception side is obtained through a transmission controller 24 at the time of starting transmission. The error power between the output obtained by inputting a certain number of random data to the inverse DCT circuit and an ideal calculated value is used as error information of the inverse DCT circuit. Error information 211 is inputted to a refresh number setting circuit 201, and the number of times of refresh is determined by a preliminarily determined method. Hereafter, refresh is executed at intervals of the number of times of refresh.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to digitally compressed transmission of television signals (hereinafter referred to as 1'V signals), and particularly relates to discrete cosine transform (D C'
1').

[Conventional technology]

There are several methods for compressing and digitally transmitting image signals, especially 1'V signals.

Among these methods, one of the methods with good compression efficiency and large actual IA burst is motion compensated interframe coding using L)CT.

A general motion-compensated interframe coding device will be described below with reference to FIGS. 2 and 3.

'1'' signal manually generated by camera 11 is A/U converter 12
The signals are digitized by the color luminance separation and rearrangement circuit ('l'l)M) 13 and rearranged in a predetermined order by color and M degrees.

The rearranged signals are divided into 6 pixels (8 pixels horizontally and 8 lines vertically).
Each block of 4 pixels is read out, and the subtracter 14
Then, a difference is calculated between the predicted signal 31 and each corresponding pixel in the block, resulting in a difference signal 32. The prediction signal 31 is
It is created from the most recently encoded and transmitted frame. The frame encoded and transmitted immediately before is stored in the frame memory 20, and the motion detection circuit 22 determines the direction and magnitude of the motion of the block to be encoded from the previous frame to the current frame. 36) is detected.

The motion vector 36 is encoded by the lJJ'N length encoding circuit 23 and transmitted. The variable delay circuit 21 reads pixels corresponding to the block to be encoded from the previous frame based on the motion vector 36, and uses this as a prediction signal.

The difference signal 32 is subjected to L)CT and converted into a fit child.

The quantized signal is encoded by the variable length encoding circuit 23, and simultaneously subjected to inverse quantization and inverse DCT to become a reproduced difference signal 33. This (No. 4 33) is added to the previous predicted No. 41 31 in the adder 19 to obtain a reproduced signal 34. The reproduced signal 34 is stored in the frame memory 20 for processing of the next frame 11.

On the receiving side, the transmitted signal kit) is decoded by the variable length decoding circuit 42 as shown in Fig. 3, and then sent as ID code 17+1.
8. 43, 44, similar to the operation of 19420.21
At 45.46, a reproduction signal (No. 4 134 is obtained).The reproduction signal 134 is subjected to color and luminance synthesis, is converted into an analog signal by an L)/A conversion 49, and is output.

If the transmitting sides 17 to 21 and the receiving sides 43 to 47 do not always operate in the same way, correct reproduced signals cannot be obtained.If both the transmitting and receiving sides operate ideally, predicted signals 31 and 131 and reproduced difference signals 33 and 133 are generated. ．． The reproduced signals 34 and 134 are exactly the same.

Although the inverse DC'L' circuits 18 and 44 must perform matrix operations on real numbers, they are actually replaced with fixed-point operations or floating-point operations using finite-length registers. At this time, there is no problem if the register length is made long enough (generally about 16 bits), but it is often not possible to make it long enough due to hardware constraints.As a result, depending on the calculation method of the inverse DCT circuits 18 and 44, Even if the same signal is input manually, the output may differ due to calculation errors, and random noise-like deterioration remains in the reproduced image on the receiving side.

'In order to remove this deterioration from the screen, a method is often used in which the switches 26 and 51 are set to the zero side once every fixed frame, and the screen is refreshed by performing intra-frame encoding. 8111 for sending and receiving
(It is also known that spreading the deterioration can be suppressed by applying the same low-pass filter to Nos. 1, 31, and 131.

However, even in this case, although the frequency of occurrence of deterioration decreases, it is necessary to perform refreshing because the deterioration that occurs is multiplied by a.

These refresh controls are performed by the refresh control circuit 25. The refresh control circuit 25
L) The counter 70 is incremented every time frame transfer No. 44 37 is received from Mn2.

The frame transfer signal 37 is transmitted once per frame as shown in FIG.
These are the same pulses. The value of the counter is compared with a fixed value 2'7 in a comparator 71, and if they match, an in-frame (id
It outputs ea5 (becomes H in FIG. 6) and at the same time clears the value of the counter 70 in synchronization with the frame transfer signal 37.

Intra-frame encoding is performed periodically with this operation,
Accumulation of errors can be prevented. Note that intraframe coding is generally several times more efficient than interframe coding. Therefore, if the refresh interval is shortened, the encoding efficiency will decrease. The amount of code to be transmitted is determined by the transmission path,
Since it is constant, if the encoding efficiency is low, either (1) reduce the number of frames to be transmitted or (2) coarsen the quantization precision.

Either the amount of code for one frame must be reduced, but in the case of (1) above, the temporal resolution deteriorates, and in the case of (2) above, the spatial resolution deteriorates.

[Problem to be solved by the invention]

In the above-mentioned conventional technology, if there is an error in the transmission/reception calculation results, refresh is performed at a predetermined station period regardless of the degree of the error, so a total of 9. In cases where the error is very small, there is a problem that the image quality deteriorates when refreshing is performed more than 8 times.The purpose of the present invention is to suppress the image quality deterioration caused by refreshing, and to provide an image and transmission method with as little image quality deterioration as possible. The purpose is to provide

[Means to solve the problem]

In order to achieve the above object, the present invention changes the refresh interval depending on the performance of the transmission destination device.

[Work W] If the performance of the transmission destination is judged to be good and the error is small, the refresh interval is lengthened, and if the performance is judged to be poor and the error is large, the refresh interval is shortened. can prevent a decline in

〔Example〕

Hereinafter, two embodiments of the present invention will be explained with reference to FIG. 1st
In the figure, 1 of the present invention is a part of 200.

At the start of transmission, the transmission controller 24 is passed through, and the reverse D on the receiving side is
Error information 211 of the C'l' circuit is obtained. The error information of the inverse DOT circuit uses the error power between the output when a certain amount of random data is applied to the inverse vCT circuit and the ideally calculated value.

The error information 211 is input manually to the refresh frequency setting circuit 201, and the refresh frequency is determined using a predetermined method. From then on, refresh is performed every time this number of refreshes are performed.

The details of the refresh frequency setting circuit 201 are shown in FIG.
vinegar. The manually entered error information 211 is latched. The latched value is input to a ROM (read only memory) 221, and the ROM 221 outputs the number of refreshes.

Table 1 shows an example of the human output of the ROM 221.

Table 1 Table 1 shows an example where the inverse OCT circuit 18 has almost no error. Reverse L) If there is an error in the CT circuit 18, the maximum value of the output value in Table 1 will be limited (for example, 4096 in the table
→ changed to 512, 1024 → 512).

In the embodiment, the error power of the inverse DCT circuit was used as the performance information on the receiving side, but instead of the error power, the average of the absolute value of the error, the worst value of the error power for each pixel, or a combination of these may be used. Similar effects can be obtained by using As performance information, you may indicate the calculation method of the inverse L) C'r circuit and the register length used for the calculation, if the inverse 1) C'f'
When using a commercially available LSI (IDC'r chip) etc. in the circuit, you may indicate the type and model number of the LSI, or send information on the manufacturer of the device without transmitting performance information. It is also possible to use this to determine the number of refreshes. Further, as in the above example, the present invention also includes setting the refresh interval using a value specified by the receiving side.

In the embodiment, the refresh interval is set only once at the start of transmission, but the present invention also includes a case where it is changed to an arbitrary time during transmission. This change may be made at regular intervals or when specified by the receiving side.

In the present embodiment, the present invention has been described using 1) a motion compensated interframe coding apparatus using CT.

In the case of an interframe coding device without motion compensation, the present invention provides
1) Other transformations other than 'C'L' (discrete sine transformation).

KL complete transformation Legendre transformation, Hadamard transformation.

Discrete Fourier transform, etc.) can also be applied. Furthermore, the present invention simultaneously generates code words, vector quantization codebooks, quantization precision, etc. during transmission and reception.

It can also be applied to refreshing when there is an error in the generation process during transmission and reception of an interframe encoding device using this.

In addition, in the above-mentioned embodiment, the interframe coding device was explained as an example, but in the case of transmitting one frame, the present invention first coarsely transmits it using L) CT or the like, and then transmits it coarsely using L) CT, etc. Transmitted by then.

This method can also be applied to an apparatus that sequentially transmits the residual difference between the reproduced frame and the reproduced frame using CT or the like. In other words, when the error in transmission and reception is large, the number of sequential transmissions is reduced, and when the error is small, the number of sequential transmissions is increased, thereby reducing the error.
# Product can be prevented.

(Effects of the Invention) According to the present invention, the number of refreshes can be set depending on the transmission destination, so the encoding efficiency can be improved.At this time, the degree of deterioration due to calculation errors in transmission and reception does not change. .

- Since the refresh interval is generally longer, it is possible to prevent a decrease in encoding efficiency due to refresh.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an image transmission device according to an embodiment of the present invention.
Fig. 2 is a block diagram of a conventional encoding device, Fig. 8 is a block diagram of a conventional decoding device, Fig. 4 is a block diagram showing details of a refresh control circuit, and Fig. 5 shows the main parts of the present invention. FIG. 6 is a block diagram showing details of a refresh count setting circuit, and is a refresh timing chart. 15-D C: 'L' 4m, l 8-...reverse L)
CTl! ! J path (sending side), 44... Reverse D C'L'
Circuit (receiving side), 25... Refresh control circuit, 2
6... Frame inside/outside selection switch, 27... Fixed value output circuit, 201... Refresh equal number setting circuit.

Claims (1)

[Claims] 1. In image transmission, a means for obtaining characteristics of a transmission partner;
means for storing the characteristics obtained by the means; means for predicting deterioration in image quality to be transmitted based on the characteristics of the other party; and a predetermined image quality deterioration reduction method or image quality deterioration removal method based on the above prediction. An image transmission device characterized by having means for selecting zero or more methods.