FPGA Implementation Of Real-time Fusion Of Infrared And Visible Light

Image fusion is a kind of method combines multi-source images of the same scene by image fusion technique so that a new image can be got which contains multi-source information. As a critical area of image fusion, fusion between infrared and visible light has been widely applied in military, security monitoring, etc. In recent years great effort has been spent on integration system of portable, low power consumption, practicability, compatibility and so on. The Infrared and visible light real-time fusion system of this thesis is proposed and developed in order to meet the requirements.For the portable and low power requirements of fusion system, overall design of the hand-held infrared and visible light image fusion system has been completed. As the core processing chip, FPGA do the fusion job of the two-source image. Simultaneously, structure of system software and algorithm of image processing both has been completed.In order to guarantee the procedure of image fusion going smoothly, the method of driving the ULIS PICO384 has been researched and realized. What’s more, processing has been done to the source image, which means preprocessing to multi-source image and image registration. Methods of preprocessing to infrared image contain Non-Uniformity Correction, Auto-gain of image. Image registration used affine transformation to fulfil the need.Practicability of portable fusion system reflected in how much useful information can be presented in the fused image. However, It needs different fusion rules for different scene. Several image fusion algorithm has been researched and fulfil in this system, which contains: the Laplacian pyramid fusion, nonnegative matrices fusion, color fusion method based on human’s characteristic. The results of the former two method were grayscale images. The result of the last method was color image.In order to reflect the compatibility of this system, Camera link format output has been applied to this system. In the process of the fusion, result image has been cached firstly, then went on coding, finally output. Meanwhile an experiment has been designed to verify the feasibility of this method.