The Real-Time Image Processing Based on DSP
Qi Chang, Sun Fuxiong, Tianshu Huang · 2005
At present, wavelet transform is gradually used in the processing of video and image. But the algorithm of 2D wavelet transform is too burdensome and complicated to process huge amount of image data in real-time. To the problem, this paper proposes a novel method to process video and image real-timely based on DSP. At the aspect of wavelet algorithm, the method changes the traditional image convolution mode, which is based on the cross filtering of rows and columns. According to the feature of lifting wavelet transform, 2D image data is firstly linearized in some scanning order to the 1D sequence. Subsequently the 1D sequence is decomposed by lifting wavelet transform. At the other aspect of hardware, the program structure of DSP is optimized to exert the operation power with the in-chip memorizer of DSP. The experiment results show that the implementation meets the real-time requirement of image coding, so is a feasible and efficient DSP solution.