A parallel line scanning method of video frequency

Abstract

The invention provides a parallel scanning method for video coefficient. The method, according to Zig-Zag scanning sequence, divides quantized coefficients of a blockette into a plurality of arrays; each array comprises m quantized coefficients; any quantized coefficient An in any array (A1, A2 ... An..., Am): if An is equal to 0, a counter value is plus 1 without outputting number pairs of range and run; if An is not equal to 0, the counter value is x, when the An is the first number which is not 0 in the array, the number pair of range and run (An, x plus n minus 1) is generated and the counter is set to 0; when An is not the first number that is not equal to 0, Au is set (u is equal to or larger than 1 and smaller than n) and is the nearest number which is not equal to 0 before An, the number pair generated by An is (An, n-u-1); number pair (level, run) is output; the process is repeated till the scanning of the blockette is finished, which only needs to consume 1 / m of the time in the prior art.

Description

technical field [0001] The invention belongs to the field of digital audio and video coding, in particular to a parallel scanning method of video coefficients. Background technique [0002] In many image coding protocols, such as H.264, AVS, etc., intra-frame image coding requires discrete cosine transform (DCT), that is, converting images from spatial expressions or spatial domains to spectral expressions or frequency domains. This is because after transformation, the image only needs a small number of data points to represent; and because the coefficients generated by DCT are easily quantized, better block compression can be obtained. After quantization, most DCT coefficients in the image matrix are truncated to zero values. When using DCT transformation, the quantized two-dimensional coefficients usually use the Zig-Zag scanning (reading) method to express the coefficients as one-dimensional, and then the number of consecutive 0 coefficients (run) and the size of the non...

AI Technical Summary

Problems solved by technology

[0003] The existing Zig-Zag scanning implementation method, as shown in Figure 1, the numbers in the circles represent the scanning order, and only one data is processed per clock cycle, so for an 8×8 quantization block, 64 clocks are required In the case of high real-time encoding requirements, such as high-definition encoding, it is difficult to achieve real-time encoding

Images