Most of recent video coding methods are based on hybrid coding, which employ motion compensation (MC) and discrete cosine transform (DCT). In these methods, prediction image is generated by MC, then prediction error image is encoded by applying DCT. But recently, video coding method using matching pursuits (MP) that is one of waveform coding techniques, in place of DCT, is suggested. The matching pursuits algorithm represents a signal approximately using a dictionary. So, if the dictionary is suitable for coded signal, MP can attain high coding efficiency. In this paper, we suggested new MP coding method that apply for prediction error image divided into sub-bands. Coding efficiency of the method can be improved by using the dictionary suitable for frequency characteristic of each sub-band.
展开▼