Researching On Image Processing Based On Dsp

With the development of computer vision and image analysis, digital image was extensively concerned in many fields. Image processing was employed abroad in national defence, industry, medical treatment and daily life.In general, image processing can be divided into three parts:image compression, image analysis and computer vision. For the image compression, JPEG2000, which is the new generation of stationary image compression standard, was issued of to replace the JPEG by the International Standard Organization (ISO). Although JPEG2000 is more complex than JPEG, the high compression ratio and Regions of Interests (ROI) encode are better than JPEG Computer vision is technology that the computer analysis the information which is got by different kinds of image system, then processes the information like human being. The base of computer vision is image analysis, that is to analysis, understand and identify the image. For the image analysis, the common method is edge detection.The development of computer science and the appearance of DSP chips provide the powerful tools for the image processing. Especially the development of DSP chips with high capability of calculation makes the embedded processing of image come true.In this thesis, we study on image compression and edge detection, which are both based on the wavelet transform. With the help of MATLAB, we make the simulation and comparison for the following edge detection algorithms:the traditional algorithm and the algorithm with wavelet transform. In order to make use of DSP chips, we change the data format of image. On the hardware platform DM642, we implement the edge detection and image compression with the wavelet transform method. In addition, according to the structure of the DSP, DMA for the transmission of image and pipelining code for the JPEG2000 are adopted to improve the speed of program.