Patent attributes
A vector coding apparatus and method includes decomposing an initial vector to obtain a sign vector and an initial absolute vector. Coding the sign vector to obtain a sign code, performing multi-level permutation-based coding on the initial absolute vector to obtain an absolute vector code, and combining the sign code and the absolute vector code. Optionally determining, before decomposing the initial vector, a characteristic codebook to which the initial vector belongs, obtaining a characteristic value of the characteristic codebook, and combining a code of the initial vector to obtain a final code of the initial vector. The performing step can include removing from the initial absolute vector any element having a value which satisfies a preset condition, constructing a new absolute vector with elements not satisfying the preset condition, and coding positions of the elements not satisfying the preset condition in the initial absolute vector to obtain a position code.
