远距离多源图像融合系统实时配准设计

为了实现全天候多波段远距离实时图像监控，设计了具有微光、红外和可见光融合的光学前端，对多源图像进行实时配准研究。在平行光轴的基础上，通过计算不同视场图像的成像视差，计算仿射变换需要的参数，采用双线性内插算法弥补红外在成像方面与可见光图像的差别，将红外图像的变换制作成查找表存储在图像处理器DM642中，系统通过硬件查找表可以快速实现不同图像的配准变换，实现同步视频的实时配准与融合。实验表明:该设计能够准确地实现多源图像的实时配准，系统经过图像配准、图像融合和伪彩变换处理后的时间约为24.3 ms，系统探测距离大于3 km。

Real-time registration for long-distance multi-source image fusion system

In order to achieve all-weather, multi-band, long-distance and real-time image monitoring, the optical front-end with low light level television, infrared and visible light system was designed, real-time multi-source image registration was investigated. Based on the parallel optical axis, the affine transformation parameters could be obtained by calculating the imaging disparity of the different field, the bilinear interpolation algorithm was used to make up the difference between infrared image and visible image in imaging, the infrared image transformation parameters were made into a look-up table stored in the image processor DM642, the system could quickly register different images through the look-up table and achieve real-time registration and synchronous fusion video. Experimental results indicate that this design can register multi-source images in real-time, the total processing time of image registration, image fusion and pseudo-color transformation is about 24.3 ms, the detection range is greater than 3 km.