Research And Design Of Visual Information Processing Sub-System Based On Reconfigurable Processor

Visual information real-time processing and display is a hot research topic.Currently, video image processing and display quality requirements are increasing. Soit is significant to design efficient video processing system. Reconfigurable processorscan complete large amount of computation using flexible reconfigurable mechanismand high computation speed. Reconfigurable processors are utilized in visualinformation processing field to improve the efficiency of the system.After studying the features and merits of the reconfigurable processor, this paperdesigns a sub-system which can display the result of image fusion from two camerasand discusses the key techniques in the processing.The reconfigurable processor parallel computation application in visualprocessing is introduced at first. Then the system architecture is given with theanalysis of data flow, hardware resource and memory space allocation. According tothe feature of rotation degrees and calculation result of the processor, the systemmodifies a three-step rotation algorithm to store data. The approach is implemented bypipeline architecture to execute real-time rotation and mosaic.In order to overcome the drawbacks of conventional histogram-based contrastenhancement methods, this paper proposes a detail preserving histogram equalizationalgorithm. It uses neighborhood data with local contrast information to generate amodified histogram and transformation functions. An adaptive parameter is applied todetermine the enhancement degree for different images.The proposed architecture can solve the problems of large scale data processingand storage. The simulation and synthesis results show that the rotation modulecircuits meet the requirement of timing and hardware consumption. The videoenhancement approach effectively enhances the contrast, alleviates theover-enhancement effect, and preserves the detail and average brightness.