GENERALIZED ORTHOGONAL TRANSFORM METHOD AND DEVICE FOR LOW RESOLUTION VIDEO DECODING

摘要

PROBLEM TO BE SOLVED: To substantially reduce a system cost by using a low resolution generalized inverse orthogonal transform means so as to convert an inverse quantization coefficient into a transform pixel. SOLUTION: A low resolution generalized inverse orthogonal transform means 230 obtains low resolution predicted error information generated through motion prediction of a video coder. First a syntax purser and variable length decoding means 210 decodes a video stream 201. An inverse quantizer 220 applies inverse quantization to a variable length decoding coefficient 211 and a frame buffer 270 stores a low resolution video image. An up-sampling means 250 retrieves a low resolution reference pixel 271, which is interpolated to generate an up-sampled pixel 251 for an inverse motion compensation means 240. The inverse motion compensation means 240 executes half pixel inverse motion compensation on the basis of the up-sampled pixel 251 to obtain a motion compensation pixel 241. An inverse quantization coefficient 221 is converted by the low resolution generalized inverse orthogonal transform means 230 and a converted pixel 231 is obtained.