JPH06253278A - Sub band picture coder - Google Patents

Abstract

PURPOSE:To attain optimum code quantity distribution even to various input pictures by sequentially revising a quantization step width of each band based on a difference between an input picture signal and a local decoding signal. CONSTITUTION:After an input picture signal is divided into 1/2 in the horizontal direction by horizontal direction sub band filters 2, 3, and vertical direction sub band filters 4-7 are used to divide the signal into 1/2 in the vertical direction, then a signal HH with high frequency in the horizontal and vertical directions, a signal HL with a high frequency in the horizontal direction and a low frequency in the vertical direction, a signal LH with a low frequency in the horizontal direction and a high frequency in the vertical direction, and a signal LL with a low frequency in the horizontal and vertical directions are outputted. Then coders 8-11 quantizes the data and the result is outputted to an entropy coder 12 and part of quantized data is locally decoded and a square of a coded error with a difference from the input signal is obtained and outputted to a control signal generator 15. The ratio of a step width of quantization of each band is revised depending on the magnitude of square mean value of the coded error to execute code quantity distribution control. Thus, optimum code quantity distribution is attained even to various input pictures.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus using a subband coding method, and more particularly to a subband image coding apparatus suitable for recording digital video information and digital video information.

[0002]

2. Description of the Related Art As an image coding method, there is a generally well-known subband coding method. This method is described in, for example, Journal of Television Society Vol. 44, No. 2 (199
0), pages 154 to 156, the input video signal is divided into a plurality of bands, and each band signal is encoded by an encoding method suitable for the property.

In the above sub-band coding method, DPCM (differential pulse code modulation) is used for coding each band.
In the case of using, the code amount control and distribution depended on the variance of the prediction error signal.

In the technique of the above-mentioned document, it is necessary to perform a process of repeatedly calculating and converging a coded value, which is inconvenient as a process of moving image coding. Further, in order to solve such inconvenience, it is conceivable to fix the code amount distribution to each band, but in this case, the code amount distribution is not adapted to the image, which causes a decrease in efficiency and image quality. It was.

[0005]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems of the prior art. Even when an encoding method different from the band is used in subband encoding, the encoder can be favorably used. An image coding apparatus capable of distributing code amount is provided.

[0006]

SUMMARY OF THE INVENTION The present invention comprises a sub-band filter for dividing an input image signal into a plurality of bands,
The encoder includes an encoder that encodes an image signal for each band based on the output of each filter, and a control signal generator that supplies a quantization step width corresponding to each band to the encoder. A quantizer for quantizing the image signal for each band based on the quantization step width supplied from the control signal generator; a local decoder for decoding a part of the output of the quantizer; An encoding error calculator that outputs an error obtained by calculating an error between the image signal for each band and the output signal of the local decoder, and the control signal generator is the encoding error. The step width of the quantizer for each band is changed based on the output of the calculator.

[0007]

According to the configuration of the present invention, the width of the quantization step of each band is changed sequentially according to the difference between the input image signal and the locally decoded signal. Therefore, it is possible to maintain the optimum code amount distribution.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a subband image coding apparatus of the present invention will be described in detail below with reference to the drawings. Figure 1
FIG. 3 is a schematic block diagram of a 4-band division type sub-band image encoding device. In the figure, 1 is an input terminal for inputting an image information signal, 2 and 3 are divisible horizontal subband filters for dividing the input image information signal into two in the horizontal direction,
Reference numerals 4 to 7 denote vertical sub-band filters that further divide the output signals of the horizontal filters 2 and 3 into two in the vertical direction. A total of four band data HH, HL, LH and LL from the vertical filters 4 to 7. (HH is a horizontal / vertical high range, HL is a horizontal high / vertical low range, LH is a horizontal low / vertical high range, and LL is a horizontal / vertical low range).

Next, 8 to 10 are adaptively switched to SQ (scalar quantization) or VQ (vector quantization) to code the band data HH, HL and LH. Is a fourth encoder which is composed of DCT (Differential Cosine Transform) and encodes the band data LL, and 12 is an addition of outputs of these first to fourth encoders 8 to 11 such as a Huffman encoder. It is an entropy encoder that encodes with information. The output of the entropy encoder 12 is output to the output terminal 14 via the buffer 13.

Reference numeral 15 is a control signal generator, which determines the first quantization step width of the lowest quantizer from the code amount of the buffer 13. Then, the quantization step widths of the quantizers in the other bands are determined from the first quantization step width at a constant ratio. In this way, the code amount control assigned to each band is performed.

The encoders 8 to 11 have the same structure when the encoders are treated as one block.
The encoder 8 will be described as a representative example. As shown in FIG. 2, the input terminal 21 for inputting the output of the vertical filter 4 is input.
A quantizer 22 for quantizing the band data HH input to the input terminal 21, an output terminal 23 for outputting the output of the quantizer 22 to the entropy encoder 12, and a quantizer 22. A local decoder 24 for locally decoding a part of the output, and a square value of a difference between an input signal of the input terminal 21 and a decoded signal of the local decoder 24, that is, a square value of an encoding error. And an encoding error circuit 25 for outputting In the figure, reference numeral 15 denotes a quantization step width control signal from the control signal generator 15, and 26 is an output terminal for the squared value of the coding error, which is connected to the control signal generator 15.

Then, the square value of the difference (coding error) between the output of each local decoder 24 of the encoders 8 to 11 and the input to the quantizer 22 is used, that is, the value of the mean square value. Update the step width ratio of the quantizer 22 in each band according to the size of
Thereby, the code amount distribution control is performed.

Further, FIG. 3 shows the control signal generator 1 of FIG.
5 is a block diagram showing another configuration of FIG. 5, in which the control signal generator 15 is divided into a first control signal generator 15a and a second control signal generator 15b, and the first control signal generator 15a is provided in the buffer 13. The code amount of the code amount 11 is controlled, and the second control signal generator 15b determines the code amount distribution of the encoders 8 to 10 based on the information of the first control signal generator 15a and the information of the encoders 8 to 10. It is configured as follows. In this embodiment, the encoder 11 is configured to encode the lowest band including the DC component, but it is obvious that the encoder of another band may be used as a reference. is there.

By the way, in both of the above-mentioned embodiments, the encoder 22 may be of any other system regardless of DCT, VQ or SQ as long as it is a high-efficiency encoding means. Further, the step width ratio may be controlled so as to optimize the SNR (Signal Noise Ratio), or may be controlled so as to emphasize the lower range by incorporating the result of the subjective evaluation. The same control can be performed by using the level difference of the step width instead of the step width ratio. Furthermore, since the input data is image data, there is no abrupt change, and unnecessary changes in control parameters lead to lower subjective evaluation. A dead zone may be provided for performing only when the difference from the value is a certain value or more.

Although the mean value of the square of the coding error is used in the above embodiment, the mean value of the absolute value of the error may be used to reduce the calculation amount, or the bandwidth may be reduced to reduce the calculation process. It is also possible to obtain the coding error information only from the regularly or randomly extracted data without having to obtain the coding error information from all the data.

In the above embodiment, the case where the divisible filter is used to divide into four bands has been shown, but it goes without saying that the present invention can be applied regardless of the filter type and the number of divided bands. Further, although the above embodiment has been disclosed on the premise of encoding within one field of the image data, it is obvious that it can be applied to an encoding method within a frame, between fields, between frames and adaptive switching thereof. Is.

[0017]

As described above, according to the present invention, the quantization step width of each band is sequentially changed. Therefore, the optimum code amount distribution is maintained even for various input images. It is possible to expect the effect of being able to significantly improve the image quality of the decoded image.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of a subband image encoding device of the present invention.

2 is a diagram showing details of the encoder of FIG. 1. FIG.

FIG. 3 is a diagram showing another embodiment of the subband image encoding device of the present invention.

[Explanation of symbols]

 2-7 Filter 8-12 Encoder 15 Control signal generator 22 Quantizer 24 Local decoder 25 Coding error calculator

Claims (1)

[Claims] 1. A subband filter for dividing an input image signal into a plurality of bands, an encoder for encoding the image signal for each band based on the output of each filter, and each band for the encoder. A control signal generator that supplies a quantization step width corresponding to the above, wherein the encoder quantizes an image signal for each band based on the quantization step width supplied from the control signal generator. , A local decoder that decodes a part of the output of the quantizer, and an error obtained by calculating an error between the image signal for each band and the output signal of the local decoder, and the obtained error is the control signal generator. And a coding error calculator for outputting to the sub-band, wherein the control signal generator changes the step width of the quantizer for each band based on the output of the coding error calculator. Image coding device.