JPH03253191A - Moving picture coding system - Google Patents

Abstract

PURPOSE:To improve prediction efficiency and to obtain stable coding picture quality by coding an input picture while varying the period of applying in-frame coding depending on the quantity of movement of the input picture. CONSTITUTION:The moving quantity of an input picture signal is detected to decide a period (n) of in-frame coding. When the moving is rapid, the period (n) is decreased and when the moving is less, the period (n) is increased. At first, a frame F1 is subject to in-frame coding and then a frame Fn is subject to in-frame coding, then the result of in-frame coding is used from the result of in-frame coding 1 and of in-frame coding 2 which is less change to apply inter-frame coding to each of frames F2-Fn-1. The series of operation above is repeated. Thus, even when there is a rapid change in a frame in future, the deterioration in the prediction efficiency is minimized to prevent the reduction in the coding picture quality and the reverse reproduction, random access and high speed reproduction are easily realized.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] The present invention relates to a moving image encoding method for encoding moving images with high efficiency.

[Prior art] Figure 3 shows, for example, Murakami's "High Efficiency Coding Technology" (Television Society Journal Vol. 1.42, No. 11 (1988),
12 is a block diagram showing the configuration of a conventional video encoder described in pages 1198 to 1204).

In the figure, (1) is the input image signal, (20) is the frame memory 1, and (3) is the interframe prediction signal (21), which will be described later, from the input image signal (1) output from the frame memory 1 (20). The subtractor for subtraction, (22) is the subtractor (
3) an inter-frame difference signal outputted from, (8) an encoder that encodes the inter-frame difference signal (21), (9)
is a quantizer, (10) is quantized data output from the quantizer (9), (11) is a variable length encoder to which this quantized data (10) is input, and (12) is this variable length encoder. A buffer memory for speed averaging connected to the long encoder (11), (13) record encoded data output from this buffer memory (12), and (14) quantized data (
10) an inverse quantizer that inputs and outputs encoded data;
(15) is a decoder that inputs this encoded data and outputs an interframe decoded difference signal (23), (23) is an interframe decoded difference signal decoded by the decoder (15), (
17) is an adder that adds this inter-frame decoded difference signal (23) and an inter-frame predicted signal (21) to be described later;
19) is the decoded image signal output from the adder (17);
(24) is the frame memory 2 that stores this decoded image signal (19), (21) is this frame memory 2 (24)
), the interframe prediction signal (25) is output from the frame memory 2 based on the image signal information of one frame ahead.
(24) is a motion compensation unit that motion-compensates the image signal within. FIG. 2 shows a coding sequence predicted from one direction in a conventional video coding method.

Next, the operation will be explained according to FIG. Input image signal (
1) is intra-frame encoded every n frames, and the frames in between are inter-frame encoded. In other words, the input image signal (1) is not subtracted by the subtractor (3) at every n frames, and the inter-frame predicted signal (21) is subtracted by the subtractor (3) for the other frames. It is converted into an interval difference signal (22). This inter-frame difference signal (22) is encoded by an encoder (8) and then quantized by a quantizer (9) to obtain quantized data (10). By passing this quantized data (10) through an inverse quantizer (14) and a decoder (15), an interframe decoded difference signal (23) is obtained. In the adder (17), this inter-frame decoded difference signal (2
3) and the interframe prediction signal (21) to obtain a decoded image signal (19). This decoded image signal (19) is stored in the frame memory 2 (24). Store the input image signal (1) of the next frame in frame memory 1 (20),
Based on this image signal, motion compensation is applied to the image signal stored in the frame memory 2 (24) to generate an interframe prediction signal (21).

By repeating the above operations, the current frame is encoded based on prediction from past frames by motion compensation, as shown in FIG.

[Problems to be Solved by the Invention] Conventional video encoding systems are configured to encode the current frame based on predictions from past frames as described above. If there is a change, prediction from the past alone is not sufficient, and the prediction efficiency deteriorates, resulting in a corresponding decrease in encoded image quality. The problem was that functions such as random access (playback from any position) and high-speed playback could not be realized.

This invention was made to solve the above-mentioned problems, and even if there is a sudden change in future frames, it is possible to minimize the deterioration of prediction efficiency and prevent deterioration of encoded image quality. The purpose of this invention is to construct a video encoding system that can easily realize reverse playback, random access, and high-speed playback.

[Means for Solving the Problems] The video encoding method according to the present invention performs intra-frame encoding 1 and intra-frame encoding while changing the period n for performing intra-frame encoding depending on the magnitude of motion of an input image. In Equation 2, the signal with less change is used for prediction and encoding.

[Operation] The video encoding method in this invention performs intra-frame encoding 1 and intra-frame encoding 2 while changing the cycle n for performing intra-frame encoding depending on the magnitude of motion of the input image.
By using the signal with less change in prediction for encoding, it is possible to improve basic prediction efficiency and obtain stable encoded image quality, which is also necessary in storage-based video encoding methods. Easily realize reverse playback, random access, and high-speed playback. In addition, in interpolation prediction, the weighting of decoded image frames in prediction is changed according to the frame number difference (between multiple decoded image frames and the predicted frame), and furthermore, multiple When the difference between the submeasurement errors is large, an adaptive process that performs prediction from the decoded image frame with the minimum prediction error,
Improve prediction efficiency.

[Example] Hereinafter, an example of the present invention will be described with reference to the drawings. 1st
The figure shows a coding sequence in a moving picture coding system according to an embodiment of the present invention.

Next, the operation will be explained. First, the amount of motion of the input image signal is detected, and the cycle n for intra-frame encoding is determined.

When there is a lot of movement, n is made small, and when there is little movement, n is made large. The operation is determined by n
This is done in units of frames.

First, frame F1 is intraframe encoded (intraframe encoding l), and then frame F1 is encoded. Intra-frame encoding (
Intraframe encoding 2).

Next, frames F2 to F are processed using the one with the least change between intraframe coding 1 and intraframe coding 2. -3 frames are interframe encoded.

That is, Fl and F2, F, and F. (i=2゜...,
n-1) is obtained, and the smaller one is used to perform interframe encoding similar to the conventional method.

Encoding is performed by repeating the above series of operations.

[Effects of the Invention] According to the present invention as described above, while changing the period n of intra-frame encoding depending on the magnitude of motion of the input image, using the i-th and i+n-th intra-frame encoding, Since the configuration is such that frames from i+1st to i+n-1th are interframe encoded, basic prediction is possible by making predictions based on past and future frames according to movement, compared to predictions based only on the past. Efficiency can be improved and stable encoded image quality can be obtained.Furthermore, the reverse playback required for storage-based video encoding is possible by reversing the readout order of recorded encoded data from normal playback, allowing for random access. can be easily realized by starting reading from a frame that has undergone intra-frame encoding, and by reading only frames that have undergone intra-frame encoding, high-speed playback can be easily achieved.

In addition, in prediction, the weighting of decoded image frames in prediction is changed according to the difference in the number of frames between multiple decoded image frames and the predicted frame, and multiple child measurements individually predicted from multiple decoded image signals are When the error difference is large, prediction efficiency can be improved by adaptive processing that performs prediction from the decoded image frame with the minimum prediction error.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of the operation of a video encoding method according to an embodiment of the present invention, FIG. 2 is an explanatory diagram of the operation of a conventional video encoding method, and FIG. 3 is a diagram showing the configuration of the video encoding method. It is a block diagram. In the figure, (1) is the input image signal, (3) is the subtracter,
(8) is an encoder, (9) is a quantizer, (10) is quantized data, (11) is a variable length encoder, (12) is a buffer memory, (13) is recorded encoded data, (14) is an inverse quantizer, (15) is a decoder, (17) is an adder,
(19) is the decoded image signal, (20) is the frame memory 1
, (21) is the interframe prediction signal, (22) is the interframe difference signal, (23) is the interframe decoded difference signal, (2
4) is a frame memory 2, and (25) is a motion compensation unit. In addition, in the figures, the same reference numerals indicate the same or equivalent parts.

Claims (3)

[Claims] (1) Intraframe encoding is performed at n-frame periods (n is an integer of 1 or more), and interframe encoding is performed using unidirectional or bidirectional prediction from the intraframe-encoded frame for the frames in between. A moving image encoding method comprising: a moving image encoding method that includes a function of detecting the amount of motion and performs encoding while changing a cycle n for performing intra-frame encoding according to the movement of an image signal. (2) Encoding is performed while changing the period n of intra-frame encoding in units of n=2k (k is an integer of 1 or more) according to the movement of the image signal. video encoding method. (3) Set the period n for intra-frame encoding according to the movement of the image signal, n=k±L (k, L are integers greater than or equal to 1, L<
2. The moving image encoding method according to claim 1, wherein encoding is performed while changing in k) units.