JPH06133304A - Picture coding system - Google Patents

Abstract

PURPOSE:To attain smooth fast feeding reproduction without increasing number of referenced frames and increasing a code quantity by arranging an independent frame coded independently, a forward prediction dependent frame coding a difference between a predicted picture formed from the independent frame and the independent frame itself at an equal distance with respect to time. CONSTITUTION:Arrows in figure depict the relation of dependence of reference in the case of coding, independent frames I and forward prediction dependent frames P are arranged at an equal distance timewise, e.g. three frames each, and bi-direction prediction dependent frames B are arranged to remaining positions. When only the independent frames I and the forward prediction dependent frames P are selected and decoded by a variable length decoder, a fast feed reproduction picture is obtained. That is, since only the independent frame I only is used to be referenced at the decoding of the forward prediction dependent frame P, number of reference frames is less and the code quantity of the forward prediction dependent frames P is not much, lots of the forward prediction dependent frames are inserted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is formed of continuous frames of image signals that are continuously input from independent frames that are independently encoded in a frame image and independent frames that are temporally preceding. Forward prediction, which encodes the difference from the prediction image, and bidirectional prediction, which encodes the difference between the prediction image formed from the independent frame and the forward prediction non-independent frame, which are temporally located before, after, or in both directions. The present invention relates to a moving picture coding system in which a non-independent frame is provided and a coding means for each frame is provided.

[0002]

2. Description of the Related Art In order to reduce the amount of code at the time of coding a moving image signal and achieve highly efficient coding, interframe predictive coding is performed by utilizing the correlation between frames of the image signal. This is because a normal moving image is quite similar between each frame, so the signal of the frame to be encoded can be predicted from the signal of the previous frame that was encoded, and the signal between the prediction result and the actual image can be predicted. Only the prediction error (residual error) is encoded.

By combining this predictive coding with orthogonal transform and variable length coding, the code amount can be further reduced. However, if there are errors or errors in decoding the prediction error, these errors may accumulate and a large difference may occur between the original signal and the decoded signal. Therefore, independent frames are set at predetermined intervals in continuous frames of continuous image signals, and these independent frames are independently coded within the frame. Then, the non-independent frame to be predictively coded forms a prediction signal based on the signals of these independent frames, and codes based on that. In such an inter-frame predictive coding method, when a certain frame is to be decoded, the data are accumulated in the past, and therefore a large number of past data are required. If you want to special playback of moving images recorded by such a method, that is, random access, visual search,
In the case of fast-forward playback, reverse playback, etc., considerable difficulties occur.

According to the invention described in Japanese Patent Laid-Open No. 2-192378, independent frames are set at regular intervals from consecutive frames of image signals that are continuously input, and these independent frames are independently set within the frame. The prediction signal of the non-independent frame between these independent frames is encoded based on the signals of the preceding and following independent frames, and these non-independent frames are predicted based on these prediction signals and their prediction error is calculated. It has been proposed to encode. In that case, regarding special reproduction, it is described that visual search and fast-forward reproduction are possible by decoding only independent frames. Further, in this known technique, in the case of predicting a non-independent frame, since the front and rear directions are uniform, reverse reproduction can be performed by the same method as normal reproduction.

[0005]

According to the above-mentioned known technique, in the case of fast-forward reproduction, only independent frames are decoded and reproduced. However, if the number of independent frames is small, the reproduced image will be chopped when fast-forwarding. Also, if the number of independent frames is large, the code amount will increase accordingly,
The benefits of predictive coding are lost. It is also not desirable that the fast forward speed of the fast forward playback be limited by the number of independent frames. On the other hand, in the case of reverse reproduction, it is inconvenient that the same amount of processing as in normal reproduction must be performed even though the same image quality as in normal reproduction is not required. Further, in the case of fast forward reproduction in the reverse direction, there is the same problem as in the case of fast forward reproduction in the forward direction. Therefore, an object of the present invention is to smooth as much as possible in a special reproduction such as fast-forward reproduction or reverse reproduction in an inter-frame predictive coding method without increasing the code amount and increasing the number of frames referred to during reproduction. An object of the present invention is to provide an encoding method capable of various reproductions.

[0006]

According to the invention described in claim 1, this object is achieved as follows. That is, in successive frames of image signals that are continuously input,
A forward prediction non-independent frame that encodes the difference between an independent frame that is independently encoded in a frame image and a prediction image that is formed from an independent frame that is positioned temporally earlier, and a forward, backward, or both directions in time. A bidirectional prediction non-independent frame that encodes the difference between the prediction image formed from the independent frame and the forward prediction non-independent frame located at
In the moving picture coding system provided with coding means for each frame, independent frames and forward prediction non-independent frames are arranged at equal intervals in time. According to the invention of claim 2, the above object is achieved as follows.
That is, in consecutive frames of consecutively input image signals, a forward frame that encodes a difference between an independent frame that is independently encoded in a frame image and a prediction image that is formed from an independent frame that is temporally preceding. A predictive non-independent frame and a bidirectional predictive non-independent frame that encodes the difference between a predictive image formed of an independent frame and a forward predictive non-independent frame that are temporally located in front, in the rear, or in both directions are provided for each frame. In the moving picture coding system provided with the coding means (1), the predictive coding of the bidirectional prediction non-independent frame is limited to the backward reference only.

[0007]

The invention described in claim 1 is suitable for fast-forward reproduction. In the case of fast-forward reproduction, only independent frames and forward prediction non-independent frames arranged at equal intervals need to be decoded and reproduced. In that case, the number of reference frames is small because the frame to be referred to when decoding the forward prediction non-independent frame is only the preceding independent frame. Since the code amount of the forward-prediction non-independent frame is not so large, a large number of forward-prediction non-independent frames can be inserted, whereby a smooth fast-forward reproduction image can be obtained. Further, in this way, since it is not necessary to reproduce all the independent frames and the forward prediction non-independent frames at the time of fast-forward reproduction, some flexibility of the fast-forward speed can be obtained. The invention described in claim 2 is suitable for reverse reproduction. For example, in the case of backward fast forward reproduction, only independent frames and bidirectionally predicted non-independent frames are decoded and reproduced. As a result, the number of frames to be referenced is small, and similarly smooth reproduction is possible.

[0008]

Embodiments of the present invention will now be described in more detail with reference to FIGS. FIG. 1 shows the entire apparatus according to an embodiment for implementing the image coding method according to the present invention. This apparatus includes an input device 101, a preprocessing unit 102, an encoder 103, a storage device 104, a decoder 105, a postprocessing unit 106, and an output device 107. The input device 101 inputs moving image signals of various formats such as Y, Cb and Cr. The input signal is converted into a format required by the encoder 103 in the preprocessing unit 102. The encoder 103 encodes this signal into a code with a minimum amount of deterioration and a small amount of data. The encoded bit stream is stored in the storage device 104 including, for example, a CD (compact disc), a DAT (digital audio tape), an HD (hard disc), and the like.

The bit stream read from the storage device 104 is decoded by the decoder 105 to form a reproduced moving image. The reproduced moving image signal is subjected to processing such as line interpolation, pixel interpolation, rate conversion, frame field conversion, and pixel aspect ratio conversion in the post-processing unit 106 according to the specifications of the output display. The image reproduced in this manner is output by the output device 107.

FIG. 2 is a block diagram showing the configuration of the encoder 103. The coding mode is roughly divided into two modes: an independent coding mode and a non-independent coding mode. In the independent coding mode, the input image is coded as it is.
In non-independent coding mode, or predictive coding mode,
The predicted image is formed by referring to the already coded image in any of the three prediction modes of before, after or these interpolations, and the difference between this predicted image and the actual image is coded. The encoder 103 includes a differentiator 2 that forms a difference between an input image and a predicted image, an orthogonal transformer 3 that reduces a code amount, a quantizer 4, and a variable length encoder 5. Independent frames can be coded only by this, but since predictive coding is performed, in addition, after the quantizer 4, an inverse quantizer 7, an orthogonal inverse transformer 8, an adder 9, and a frame memory & The predictor 10 is connected to the frame memory & predictor 10.
The output terminal of is connected to the difference unit 2 and the adder 9 via the switch 11.

In the independent coding mode, the switch 11 is open and the difference unit 2 is disabled, so that the input terminal 1
The image signal input from is sent to the orthogonal transformer 3 as it is. The input image is subjected to discrete cosine transform here, and the resulting transform coefficient is quantized by the quantizer 4, variable-length coded by the variable-length encoder 5, and the digital storage medium (DSM) or the like from the output terminal 6. Sent to. In the case of the predictive coding mode, the difference from the predicted image according to the predictive mode is formed by the differentiator 2, and this difference is transformed, quantized and coded as in the case of the coding of the independent frame. Is output.

The predictive image in the predictive coding mode is formed by the predictor 10 from locally decoded images accumulated in the frame memory & frame memory of the predictor 10. These locally decoded images are images locally decoded by the inverse quantizer 7 and the orthogonal inverse transformer 8 connected after the quantizer 4. This decoded image is stored in the frame memory 10
The prediction image formed by the predictor 10 is added by the adder 9 from the already created another locally decoded image accumulated in the above, and formed. The formed image is stored as a predicted image in the frame memory 10 and is continuously used for forming a predicted image in the future.

FIG. 3 is a block diagram showing the configuration of the decoder 105. The decoder 105 includes a variable length decoder 13, an inverse quantizer 14, an orthogonal inverse transformer 15, an adder 16, and a frame memory & predictor 17, and the configuration and operation are as follows.
Except for the variable length decoder 13, it is exactly the same as the local decoding part 7, 8, 9, 10 in the decoder 103. That is, the data input from the input terminal 12 is decoded by the variable length decoder 13, inversely quantized by the inverse quantizer 14, and inversely discrete cosine transformed by the orthogonal inverse transformer 15. This signal is a decoded image, and this image is added by the adder 16 with another image already made stored in the frame memory 17 to form a predicted image.

Next, the frame arrangement of the interframe predictive coding system will be described. FIG. 4 shows an embodiment of a frame arrangement according to the present invention. Here, some frames are numbered in the order of display. The arrows in the figure show the reference dependency at the time of encoding. Independent frames are represented by I, forward predictive non-independent frames by P, and bidirectional predictive non-independent frames by B. In the configuration of FIG. 4, the independent frame I and the forward prediction non-independent frame P are arranged at equal intervals in time, and in this example, the independent frame I or the forward prediction non-independent frame P appears every three frames. It is arranged. By selecting and decoding only the independent frame I and the forward prediction non-independent frame P in the variable length decoder 13, a fast-forward reproduced image can be obtained. The order of the reproduced images in that case is as shown in FIG. 5, which corresponds to triple speed reproduction. Also, if decoding and reproduction are performed in the order of I0, P6, I12, ..., 6 × speed reproduction is also possible.

FIG. 6 shows another example of the frame arrangement according to the present invention. Here the frame order is the same as that according to FIG. 4, but the bi-predictive non-independent frame B is
Only the backward independent frame I or the forward prediction non-independent frame P is referred to in the predictive coding. Since the reference frame of the bidirectionally predicted non-independent frame B is only one frame that is positioned backward in time, the number of frames to be held during reverse reproduction is significantly reduced.

[0016]

According to the present invention as set forth in claim 1, smooth fast-forward reproduction can be performed without increasing the number of independent frames, that is, without significantly increasing the code amount. The frame memory for storing the frames to be stored does not need to be increased so much. According to the second aspect of the present invention, the number of frames to be referred to during reverse reproduction can be significantly reduced, and smooth reverse reproduction is possible. In any case, according to the present invention, it is possible to perform smooth special reproduction by reducing the number of frames to be referenced as much as possible during special reproduction.

[Brief description of drawings]

FIG. 1 is a block diagram of an apparatus that implements an interframe predictive coding scheme according to an embodiment of the present invention.

FIG. 2 is a block diagram showing an embodiment of an encoder.

FIG. 3 is a block diagram illustrating an embodiment of a decoder.

FIG. 4 is a diagram showing an example of a frame arrangement according to the present invention.

FIG. 5 is a diagram showing a frame arrangement during fast-forward reproduction according to the present invention.

FIG. 6 is a diagram showing another example of a frame arrangement according to the present invention.

[Explanation of symbols]

 2 differentiator 3 orthogonal transformer 4 quantizer 5 variable length encoder 7 inverse quantizer 8 orthogonal inverse transformer 9 adder 10 frame memory & predictor 11 switch 13 variable length decoder 14 inverse quantizer 15 orthogonal inverse Converter 16 Adder 17 Frame memory & predictor

Claims (2)

[Claims] 1. In successive frames of consecutively input image signals, a difference between an independent frame that is independently coded in the frame image and a prediction image formed from an independent frame that is temporally preceding is calculated. A forward prediction non-independent frame for encoding and a bidirectional prediction non-independent frame for encoding the difference between an independent frame located in front, rear, or both directions in time and a prediction image formed from the forward prediction non-independent frame are provided. In the moving picture coding method provided with coding means for each frame, an independent frame and a forward prediction non-independent frame are arranged at equal intervals with respect to time. 2. In successive frames of image signals that are continuously input, a difference between an independent frame that is independently coded in a frame image and a prediction image formed from an independent frame that is temporally preceding is calculated. A forward predictive non-independent frame to be encoded, and a bidirectional predictive non-independent frame to encode a difference between a predictive image formed from an independent frame and a forward predictive non-independent frame positioned temporally before or after or in both directions, are provided, respectively. In the moving picture coding method provided with the frame coding means, the picture coding method is characterized in that the predictive coding of the bidirectional prediction non-independent frame is limited to only the backward reference.