Real-Time Image Processing System Based On ARM And DSP

With the fast development of modern information technology, the researches and applications of image processing, especially real-time image processing catch great interests from all over the world. The ITU-T and ISO/IEC work together to constitute a new H.264 video compress standard, and it provides significantly better video compression efficiency together with the enhanced ability of error resilience and network adaptation. For example, at a same vision quality level, H.264 can improve 50% compress efficiency to H.263.The excellent performance of H.264 is at the cost of the increase of computational complexity. In order to widely apply H.264 standard to more areas, TI DSP processor specially designed for the DM Series, with the following characteristics: high-speed computing capability, high-speed data transmission, high-capacity data storage, the flexibility of dedicated video interface. However, the relatively simple interface and control capabilities of DSP chip greatly limit their application in a lot of areas. The project complete the real-time image processing system for the combination of TI DM642's outstanding image processing capability and Samsung ARM9 chip S3C2410 mainframe control.In view of certain areas of the image real-time monitoring and data backup requirements, in my design, the captured images from camera are processed in two ways: First, the camera image processing in DSP via H.264 compression, and ARM read data through the HPI of DSP, and then make storage to USB device. Second, the compressed bit stream transmits through the network interface integrated in DM642 chip to PC machines. PC decodes the data sent over by the video decoding process to obtain images of the camera.This paper discusses in detail the design and implementation of the two functions, and analyzes the results.