JPS6340392B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an adaptive predictive coding method for band compression coding a television signal.

Conventional interframe coding methods such as conditional pixel replenishment method and composite difference method use interframe difference coding with the pixel at the corresponding position one frame before, or pre-predictive coding of the interframe difference. Band compression of television signals has been carried out. With these methods, when there is little movement on the screen, it is possible to transmit a television signal with good image quality using a small amount of information, but when the movement is rapid, the prediction error becomes large, and the compression efficiency is poor, resulting in poor image quality. Quality cannot be maintained. In order to solve this drawback, the following techniques are used in the conventional technology.
That is, when a frame of a television screen is divided into blocks each containing a suitable number of n.times.n pixels, the movement of the image can be regarded as mutually parallel movement when viewed locally within this block. Therefore, considering the motion of moving a pixel to the right per frame as shown in Fig. 1, the predicted value Z^ _ij of pixel Z _ij is calculated as Z^ _ij = y _ia , _j
By setting Z^ _ij = y _ij in the conventional method
The prediction efficiency can be improved compared to normal interframe coding. However, with this method, it is necessary to send information indicating the image movement to the receiving side for each block separately from the prediction error, and in addition, when the image movement is smaller than one pixel per frame, the movement of non-integer pixels per frame is necessary. It had the disadvantage that the prediction effect was poor when .

In order to solve these shortcomings of the prior art, the present invention encodes the television signal using an adaptive prediction method that follows the movement of the television screen and does not require motion information. The present invention provides an adaptive encoding method that can improve the compression rate and the image quality.

The present invention will be explained in detail below with reference to the drawings.

In the present invention, as shown in FIG.
Block B _x to be encoded consisting of multiple pixels of F ₀
and the current field F ₀ to which the immediately preceding reference block B _r belongs, and the previous field F ₁ and previous field F ₂ sequentially adjacent to the current field F ₀ are spatially different from the reference block B _r on the scanning line. A reference block _Bs having the same shape as the reference block _Br is set at a position corresponding to the reference block Br, and the reference block _Bs and its reference block _Bs are set in the previous field _F1 in units of at least one pixel, for example, vertically, horizontally, and vertically on the field. Between each block obtained by moving the reference block B r and the reference block B _r , and within the previous field F ₂ , the reference block B _s and its reference block B _s are moved in the same direction as the movement and twice as large. The intra-block average value of the squared error (or absolute value error) of the inter-field difference signal and the squared error of the inter-frame difference signal between each block obtained by moving by 2D and the reference block B _r . (or absolute value error) within the block. A set of blocks B _c1 for which the value of this sum is the minimum,
From the position of B _c2 , the amount of movement of the reference block B _r in the current field F ₀ is determined, and then two of the inter-field difference signals of this set of blocks B _c1 and B _c2 are calculated.
Compare the squared error and the squared error of the interframe difference signal,
Define the smaller block as similar block B _c0 . Prediction is performed on the next adjacent block B _x to be coded on the scanning line of the reference block B _r in the current field F ₀ , and the predicted value is
The next adjacent predicted block B _p to the similar block B _c0 is used.

That is, as shown in FIG. 2, the amount of movement of the reference block B _r of the current field F ₀ containing the pixel Z _ij is determined using the previous field F ₁ and the previous field F ₂ .
Furthermore, by comparing the squared errors, a similar block B _c0 is selected, and the predicted value of the pixel Z _ij ^* at the same position as the pixel Z _ij in the adjacent block in the current field F ₀ is selected. pixel Z _ij
Select the pixel Y _ij ^* located at the same position as .

For adjacent blocks in the same field,
Using this prediction method, which focuses on the fact that the amount of motion with respect to the previous field F ₁ and the previous field F ₂ is almost the same, detects the amount of motion using between frames and between fields, which eliminates motion errors due to noise. Detection can be avoided, and there is no need to send motion information to the receiving side, making it possible to enhance the compression effect even for screens with rapid movement. Furthermore, since prediction is adaptively performed between frames and between fields, by using interfields that are strong against movements of non-integer pixels, the compression effect can be improved even for movements that cannot be tracked between frames.

FIG. 3 shows an embodiment of the present invention, in which 1 is a signal input terminal, 2 is a memory, 3 is a blocking circuit, and 4 is a block circuit.
is a block memory, 5 is a delay circuit, 6 is a block memory, 7 is a motion detection circuit, 8 is a frame memory, 9 is a similar block detection circuit, 10 is a prediction circuit, 11 is a subtracter, 12 is a quantizer, 13 is a 14 is a buffer memory, and 15 is a signal output terminal.

First, a digitized television signal is inputted from a signal input terminal 1 and stored in a memory 2 for one field. In the blocking circuit 3, (n
A block consisting of xn) pixels is created, similar blocks are detected, and stored in the block memory 4 until a predicted value is determined. After the determination, the delay circuit 5 applies a delay to the current block in order to determine the predicted value of the next block, and the current block is stored in the block memory 6. The motion detection circuit 7 calculates the intra-block difference average based on the output of the block memory 6 and the pixel values of the previous field and previous field stored in the frame memory 8, thereby detecting the movement of the block. is detected both between frames and between fields. Similar block detection circuit 9 detects similar blocks by comparing the inter-field intra-block difference average value and the inter-frame intra-block difference average value. The prediction circuit 10 determines the previous field or the block to the right of a similar block in the previous field as the predicted value. The predicted value is subtracted from the corresponding pixel stored in block memory 4, and the result is quantized in quantizer 12. The quantized output is locally decoded by adding it to the predicted value from the prediction circuit 10 in an adder 13,
It is stored in the frame memory 8. At the same time, the signal is guided to the signal output terminal 15 via the buffer memory 14.
It is encoded into a desired code format by an encoder (not shown).

FIG. 4 shows a specific example of the motion detection circuit 7 of FIG. 3, in which A1 is an arithmetic circuit, A2 is a previous field block memory, A3 is a previous field block memory, A4 is a comparison circuit, A5 is a memory, and A6 is a comparison circuit. is the previous field address designation circuit, A7 is the previous field address designation circuit, A8 is the motion amount memory, and A
9 is a switch, 8,100,102,10
3 is the same as in FIG.

First, pixels a ₁₁ , a ₁₂ , . . . included in the block of the current field sent from the block memory 6.
...a _oo are _the pixels b 11 , b ₁₂ , b ₁₁ , c ₁₁ , c included in the block of the previous field and previous field stored in the previous field block memory A2 and the previous field memory A3. ₁₂ ...... _coo , the arithmetic circuit A1 calculates the inter-field intra-block difference average value and the inter-frame intra-block difference average value. The previous field address designation circuit A6 and the previous field address designation circuit A7 designate the address of the block stored in the previous field block memory A2 and the previous field block memory A3 to the frame memory 8. The amount of addressing in the previous field is always twice as large as the amount of addressing in the previous field.
When the calculation is completed in the arithmetic circuit A1, another address specification is performed. Comparison circuit A4, memory A5, and switch A9 determine the amount of address specified when the sum of the intra-frame intra-block difference average value and the inter-field intra-block average difference value is the minimum, which is the previous field of the block of the current field, and the previous field of the previous field. The motion amount is stored in the motion amount memory A8.

FIG. 5 shows a specific example of the similar block detection circuit 9 shown in FIG. 3, in which B1 is an arithmetic circuit, B2 is an arithmetic circuit, B3 is a previous field block memory, B4 is a previous field block memory, and B5 is a comparison circuit. , B6 is a previous field similar block addressing circuit, B7 is a previous previous field similar block addressing circuit, B7 is a previous previous field similar block addressing circuit, and 8,101,10
3,104 is the same as in FIG. The previous field similar block address designation circuit B6 and the previous field similar block address designation circuit B7 designate an address corresponding to the amount of motion detected by the motion detection circuit 7 to the frame memory 8,
In the previous field, a block of addresses twice the size of the previous field is stored in the previous field block memory B3 and the previous field block memory B4. The arithmetic circuits B1 and B2 calculate the inter-field intra-block difference average value and the inter-frame intra-frame difference average value, and the comparator circuit B5 determines the magnitude relationship between the blocks and the blocks of the previous field and the previous field. Block is selected.

FIG. 6 shows a specific example of the prediction circuit 10 shown in FIG. 3, in which C1 is a block address shift circuit, C2 is a block memory, and
07 is the same as in Figure 3, B101, B10
2, B103 is the same as in FIG. The 1-block address shift circuit C1 shifts the address of a similar block one block to the right, specifies the address of the block adjacent to the right of the similar block in the frame memory 8, and stores the block at the specified address in the block memory C2. be done.

As explained above, by using an adaptive prediction method that follows motion and does not require motion information, the present invention can configure a coding method with high prediction efficiency and excellent image quality. It can be applied to high-quality, high-compression encoding methods such as

[Brief explanation of the drawing]

Figure 1 is a diagram for explaining the conventional method effective for interframe prediction when the image movement is an integer number of pixels per frame, and Figure 2 is a diagram for explaining the interframe and interfield motion for moving images according to the present invention. A diagram for explaining the principle of prediction, FIG. 3 is a block diagram showing a configuration example of an inter-frame and inter-field band compression encoding device according to the present invention, and FIG. 4 shows motion used in the embodiment of FIG. 3. 5 is a block diagram showing an example of the configuration of a similar block detection circuit used in the embodiment of FIG. 3; FIG. 6 is a block diagram of a similar block detection circuit used in the embodiment of FIG. 3. FIG. 3 is a block diagram showing an example of the configuration of a prediction circuit. F ₀ ...Current field, F ₁ ...Previous field,
F ₂ ... Previous field, B _x ... Block to be encoded, B _r ... Reference block, B _s ... Standard block, B _c1 , B _c2 ... Corresponding block, B _c0 ... Similar block, B _p ...Prediction block.

Claims (1)

[Claims] 1. For encoding a television signal, a current field to which a block to be encoded including a plurality of pixels and a reference block immediately before it belongs, and a previous field and a previous field sequentially adjacent to the current field. A reference block having the same shape as the reference block is set at a position spatially corresponding to the reference block on the scanning line in each of the previous fields, and the reference block and the reference block are arranged in at least one pixel unit within the previous field. The inter-field error is calculated for each pixel between each block obtained by the movement and the reference block, and the reference block and the reference block are measured in the same direction as the movement direction, but also in the second direction. The inter-frame error is calculated for each pixel between each block obtained by moving the block by twice the size and the reference block, and a set is determined that minimizes the sum of the inter-field error and the inter-frame error. The block for which the smaller error is obtained by selecting corresponding blocks and comparing the inter-field error of one block of the selected set of corresponding blocks with the inter-frame error of the other block. is selected as a similar block, and each pixel included in the block to be encoded is determined by each pixel included in the next adjacent block of the similar block on the scanning line in the previous field to which the similar block belongs or in the previous field. An adaptive predictive coding method for television signals, which is characterized by predicting the prediction error and encoding the prediction error.