面向SoC的红外图像融合算法

Algorithms of Infrared Images Fusion for SoC

本文针对红外图像和可见光图像的特点,研究了像素级的三类图像融合算法。选择适合在硬件上实现的拉普拉斯金字塔图像融合算法,针对算法中的滤波和融合规则进行了改进,并提出了一种评价体系。然后在ARM9的S3C2440平台上完成了系统的设计,给出了系统实现的结果,最后采用评价体系对几种融合方法的结果进行了定量比较,确定在本系统中最适合的方法是改进的多融合规则的拉普拉斯金字塔图像融合算法。融合结果较好的符合了人的视觉特性。

The paper studied three kinds of image fusion algorithms in point to characteristic of infrared images and visible images based on pixel. According to the hardware of the system, the Laplacian pyramid image fusion algorithm was chose and improved such as filtering and the rule of fusion. Then, the evaluation system was put forward. Design of the system was done in the $3C2440. The results of the fusion were listed. At last, the results of several~ fusion methods were particularly compared by the evaluation. The Laplacian pyramid image fusion algorithms of some fusion rules was adapt to the system. The effect of the fusion can accordance with the visual characteristics of people.