An Analytic Transform Kernel Derivation Method for Video Codecs

In the standardization of versatile video coding (VVC), discrete cosine transform (DCT)-2, sine (DST)-7, and DCT-8 are regarded as primary kernels. However, DST-4 DCT-4 can also be considered kernels instead using DST-7 owing to their effectiveness in smaller resolution test sequences. To implement these different block size kernels, a considerable amount memory has allocated. Moreover, consumption store is major issue standardization. address this problem, common sparse unified matrix concept introduced study, where any kernel obtained after some mathematical operations. The proposed saves approximately 80% static by storing only few elements for DCT-2, DST-7, DCT-8. Full-required derived stored generated unit-element matrices permutation matrix. required 1648 8180 elements, each with 8-bit precision. defined composed two parts: DST-3 grouped used derive points DCT-2 new technique grouping introduced, which shows relationship between rows various sizes. supports fast algorithm implementing method “one group one feature” principle. simulation was conducted VTM-3.0 reference software under conditions. result all intra (AI) configuration Y = 0.00%, U −0.02%, V 0.00% an encoding time 100%, decoding 100%. Similarly, results random access (RA) −0.01%, 0.09%, 0.06%, times 101% respectively. low delay B (LDB) 0.01%, 0.08%, −0.27%,