Dim Light And Infrared Video Fusion Image Processor

Image fusion technology, as an important part of the multi-sensor information fusion, has become more and more popular in recent years. Among them, the image fusion of low-level-light and infrared has become a research hotspot. In modern combat, the advantage of being able to see further and better under low light conditions is of paramount importance in determining the outcome of a battle. Low-level-light image has high resolution and quality while it is easily influenced by the environment. Thermal infrared sensors can provide high contrast image and be operated normally in all weather while the infrared image has low resolution. If we could fusion these two kinds of image by some methods, the observation ability of the night vision system would be greatly improved.This thesis mostly focuses on the research of the image processor of the low-level-light and infrared image fusion system. Starting from the basic theory, the thesis introduces the development of fusion systems at home and abroad, and studies the source and contrasts the characteristics of low-level-light and infrared images. It also summarizes some common image fusion algorithms and methods used to pre-process images and estimate the fusion images. The thesis also emulates some algorithmic approaches with MATLAB and chooses the best one for the system. Then, the thesis gives a brief introduction of the hardware part of the self-developed TMS320DM642-based image processor system. The thesis uses the CCS2 to software test the devices on board, including the DSP, SDRAM, FLASH, serial port and video ports, and to burn the FLASH. It uses image enhancement and image registration based on simple affine transformation model in pre-processing, selects weighted pixel average algorithm in image fusion and uses the DSP/BIOS to build a framework of procedures to complete the multi-threaded system, which can achieve real-time image output requirements in the experiments indoors and outdoors.