Adaptive selection model for detecting zero-quantized discrete cosine transform coefficients in video coding

An adaptive selection approach for advanced video coding that reduces the number of unnecessary discrete cosine transform (DCT), quantization (Q), inverse quantization (IQ), and inverse DCT (IDCT) computations is proposed. An adaptive selection approach is used to detect zero-quantized DCT (ZQDCT) coefficients since certain zero-quantized DCT coefficients of the prediction residual block are difficult to identify. The presented algorithm detects ZQDCT coefficients according to the corresponding frequency positions before implementing DCT. Therefore, redundant DCT, Q, IQ, and IDCT procedures are removed. To carry out the proposed algorithm, three sufficient conditions are assumed for deriving eight prediction modes, which are employed to select various types of DCT, Q, IQ, and IDCT implementations. The computed results indicate that the proposed algorithm is capable of reducing the number of DCT, Q, IQ, and IDCT computations compared to those required by related methods, while retaining the coding performance of the original encoder.