Based On TMS320DM642 Visible Light And Infrared Video Image Fusion System Realization

This paper mainly studies the realization of the visible and infrared video fusionsystem. The purpose is to provide the feasibility of the automotive night vision systemwhich has reliability, good vision and quick processing speed. Image preprocessingand target extraction, image fusion and the image fusion system based onTMS320DM642are three important parts of this paper. And the main content is asfollows:(1) Image preprocessing and target extraction section describes the generalpreprocessing algorithms, including the color space introduction, image enhancement,target detection and so on. In the image enhancement, the S-Retinex brightnessadjustment for night-vision image is presented. The experimental results show that itcould retain the essential light and shade information, adjust the global brightness andavoid the color distortion by the basic Retinex algorithm.(2) According to the comparisons within the abundant experiments, the SUSANalgorithm is used in the target detection, which has good antinosie ability, lowercomplexity and better Real-time.(3) In image fusion section, the color fusion methods based on the brightness andcontrast transfer and the fusion methods based on the color transfer, provides a colorfusion method based on infrared target detection which combines the S-Retinexbrightness adjustment for night-vision image and he SUSAN algorithm. According tothe experimental results, this method is of good Real-time and good visualobservation.(4) The image fusion system based on TMS320DM642section, first introducesdifferent video port function modules in detail, then on the basis of those modulespresents the thought and process of the video captured program, the video displayedprogram and the main program. Though there are some technological problems andthe limitations of experimental conditions, the result still shows it has better visualobservation than the weak-light visible image.