Research And Implementation Of An Embedded Real-time Parallel System Of Image Processing

Infrared image processing is one of the key technology of infrared imaging guidance systems. It has three traits in common: complicated tasks, huge computational burden, high performance of real-time. So we have to construct the specially application-oriented embedded system software according with requirements of image processing in infrared imaging guidance. Based on multi-C6000DSPs support real-time infrared image processing system, the thesis focuses on the research of designing and implementing the parallel operating system software for infrared image processing. After analyzing the characteristic contained in infrared image processing and technology necessity in software system, the thesis discusses the overall design scheme and the main framework of software system based on micro-kernel structure and hierarchical model, according to the concept of embedded real-time system and distributed systems. In the following chapters, implementation of some functions, including the modules in micro-kernel, real-time communication between DSPs, and the process of system's boot and initialization, are introduced. When discussing real-time communication, the thesis describes data communication, mutex between nodes and memory sharing in this system. Then, this thesis analyzes system reliability. And it discusses the fault-tolerant especially, taking account of two instances of inspect and recovery: beginning and running time. At last, some key modules performances and application performance of the software system are tested, and the test results are also given. The results show that the whole software system can meet the demands of real-time infrared image processing.