JPS6041915B2 - Image signal encoding processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image signal encoding processing method and an image signal band compression processing method that has a simple circuit configuration and is capable of high compression.

FIG. 1 shows an example of a conventional orthogonal transform encoding processing method using prediction.

On the sending side, the process is as follows. That is, a video signal from an imaging device such as a camera is first converted into a digital signal by the A/D converter 2. The output signals are grouped into N pixels each to form a block. The Hadamard transformer 3 performs an N-order one-dimensional Hadamard transform on the block. The subtraction circuit 4 compares the Hadamard transform output with the result predicted by the predictor 10 based on the Hadamard transform output one line before. Then, the prediction error signal is converted to the sequence (Seq
The quantization and encoding corresponding to the quantization and encoding are performed by the quantization and encoder 6'. However, in this case, prediction is performed between conversion outputs of the same sequence. To do this, the number of blocks in one line must be an integer number, and as a function of the Hadamard transformer 3, the time-series input of the digitized image is divided into N blocks each. It is necessary to perform Hadamard transformation and convert the transformation output into a time series and output it. Sequence refers to the number of times each row vector of the Hadamard matrix is encoded. On the receiving side,
The Hadamard transform output is reproduced based on the adder circuit 12 and the predictor 14, and after being inversely transformed by the Hadamard inverse transformer 15, the image signal is reproduced by the D/A converter 16 (
(See Patent No. 884496, Arisawa, Kanaya: Agmar Transform Coding Transmission System).

In the figure, 8 represents an adder circuit, and 9 and 13 each represent a delay circuit. However, in the case shown in FIG. 1, since individual fixed length codes are transmitted, the code length is determined by the number of bits required for a block that is difficult to compress, making it difficult to achieve high compression.

An object of the present invention is to highly compress an image signal by performing an adaptive encoding process in which the number of encoding bits is changed depending on the properties of an orthogonal transform block in an orthogonal transform encoding processing method using prediction. .

The drawings will be explained in detail below. FIG. 2 shows an embodiment of the present invention, in which an image input from an image input terminal is converted into a digital signal by an A/D converter 2. In FIG.

The digital signal XN is passed through the Hadamard transformer 3 to obtain the Hadamard transform output YN. This Atamal conversion output Y
N is supplied to one input of a subtraction circuit 4, while the output EN of this subtraction circuit 4 is supplied to a mode selector 5 and a quantizer 6. The mode selector 5 selects one of the two modes based on the output EN of the subtraction circuit 4, controls the characteristics of the quantizer 6, and controls the encoder 7. A transmission line code is created from the output Z9 of the converter 6 and the output of the mode selector 5, and is sent to the transmission line at a constant speed.
The output ZN of the quantizer 6 is supplied to one input of an adder circuit 8, and the adder circuit 8 supplies its output to a predictor 10 through a delay circuit 9 which provides a delay of one horizontal scanning interval. The predictor 10 then supplies its output ■N to one input of the subtraction circuit 4 and one input of the addition circuit 8. On the other hand, on the receiving side, the decoder 11 processes the transmission path code. Regenerate the signal ZN from the number.

The signal ZN is supplied to one input of an adder 12, and the output of the adder 12 is supplied to a delay circuit 13 similar to the delay circuit 9 and to a Hadamard inverse transformer 15. The output of the delay circuit 13 is supplied to a predictor 14 similar to the predictor 10, and the output of the delay circuit 13 is supplied to a predictor 14 similar to the predictor 10. The output of instrument 14 is fed to one input of adder 12. The output of the Hadamard inverse converter 15 is sent to the D/A converter 16
The image output 17 is reproduced.

Next, the operation of this method will be explained. The input signal is subjected to one-dimensional Hadamard transformation using an N×N Hadamard matrix, and the transformation output is defined as .

Next, let's take the conversion output of the previous line and calculate its prediction? The selector 5 selects the absolute value 1ei1 (1=0 to N-1) of the predicted error signal and a preset threshold value Ti(1=0 to N-1) for each sequence component. 0 to N-1) and select one of the following four encoding modes.

Nakoi corresponds to sequence. Coding mode [1] This is a case where all components of the prediction error signal are less than or equal to a threshold value, and in this case, all components are not encoded.

Therefore, each component of the output ZN of the quantizer 6 is 10.
[■] In modes other than [1], all prediction error signals from N/4 to (N-1) are below the threshold, and from 0 to (N
/4-1) Encode only the components. [■] Mode [1
], [■], all prediction error signals from N/2 to (N-1) are below the threshold, and in this case, from 0 to (N/
2-1) Encode only the components.

[■] A case where even one of the prediction error signals from N/2 to (N-1) exceeds the threshold, in which case all components are encoded.

In other words, the encoding mode is based on the prediction error signals arranged in sequence (corresponding to prediction error signals ordered from low frequency components to high frequency components) that are simultaneously below a threshold from sequence N-1 (from high frequency components). The selection is based on the number of items, mode [1] is N items, mode [■
] is selected when the number is 3N/4 or more and less than N, mode [■] is selected when the number is less than N/2, and mode [■] is selected when the number is less than N/2.

When encoded using the above mode, the number of bits required for each block is the mode code (2 bits) plus the number of focuses required for encoding the component to be transmitted. Needless to say, in the orthogonal transformation signal domain, the orthogonal transformation signal output changes depending on the design of the image, and in the case of a fine design, signals corresponding to high frequency components are output, and in the case of coarse designs, only low frequency components are output. There are characteristics such as not having . It can be considered that the above encoding skillfully utilizes this feature. In the case of the conventional system shown in FIG. 1, a large compression ratio could not be obtained because the block, which is the most difficult to compress, was transmitted only in mode (2) due to fixed length code transmission.

On the other hand, the method shown in FIG. 2 has the advantage that it can be highly compressed because it can transmit data in mode I or mode II in quiet areas such as flat parts of the image. Although Hadamard transform is used as the orthogonal transform in FIG. 2, it goes without saying that other orthogonal transforms such as Fourier transform and cosine transform can also be used.

In this case, the converted outputs are arranged in order from low frequency to high frequency components. If a certain number of high-frequency components of the prediction error signal are simultaneously equal to or less than the threshold value, the number of modes will not be 4, but any number of modes is possible. According to the above configuration, the number of bits can be allocated to each component of the prediction error signal according to the characteristics of the image, such as the flat part and the limbal part of the image.

In other words, in mode I, only the mode sign is present, and in mode ■, it is 1.
/4 component, and in mode (2), it is sufficient to transmit 1/2 component. The number of bits to be transmitted can be significantly reduced compared to the conventional case where all components are transmitted. Therefore, the bandwidth of the transmission path is reduced accordingly. Furthermore, compared to the conventional configuration, high compression can be achieved with the addition of a small number of circuits such as a mode selector.
It will be obvious that the present invention can be used not only for transmitting images, but also for storing image data in a memory as digital information.

[Brief explanation of the drawing]

FIG. 1 shows an example of a conventional image signal transmission system, and FIG. 2 shows an embodiment of the present invention. In the figure, 1 is an image signal input terminal, 2 is an A/D converter, 3 is a Hadamard converter, 4 is a subtraction circuit, 5 is a mode selector, 6
is digitizer, 6″ is digitizer and encoder, 7 is encoder, 8
is an adder, 9 is a delay circuit, 10 is a predictor, 11 is a decoder, 12 is an adder, 13 is a delay circuit, 14 is a predictor, 1
5 represents a Hadamard inverse transformer, 16 a D/A converter, and 17 an image output terminal.

Claims (1)

[Claims] 1. In an image signal encoding processing method in which an image signal is orthogonally transformed to obtain a transform output, and the transform output is predicted from the transform output at least one horizontal scanning interval before in a frame to obtain a prediction error signal, the above-obtained The prediction error signal is compared with a preset threshold, and among the prediction error signals ordered from low frequency components to high frequency components, the number of prediction error signals that are less than or equal to the threshold is read simultaneously from the high frequency component, and then A mode selection unit is provided that divides the prediction error signal into a mode in which the entire prediction error signal is encoded, a plurality of modes in which a part of the prediction error signal is encoded, or a mode in which no encoding is performed at all, and the above-mentioned method is selected based on the selected mode. An image signal encoding processing method characterized by encoding a prediction error signal.