基于SIFT特征匹配的监控图像自动拼接

Automatic mosaic of surveillance images based on SIFT feature matching

针对不同摄像头的监控图像,提出了一种优化的SIFT特征匹配的监控图像自动拼接方法。在图像整合方面,通过高速提取SIFT特征描述符并进行稳定精确匹配,利用改进RANSAC算法去除错配,从而确定待拼接图像之间的变换参数;在图像融合方面,有效消除了颜色和光照差异,最终实现自动的无缝拼接系统。实验结果证明该方法对重叠区域小、形变大、有运动遮挡和噪声的监控图像有较完美的拼接效果。

In this paper; an improved Scale Invariant Feature Transform （SIFT） feature-based approach for efficient automatic mosaic surveillance images was proposed. In the image registration step, SIFT Feature was extracted efficiently and reliable matching was carried out, and Random Sample Consensus （RANSAC） was improved to guarantee stability and decide the transformation parameters of the images stitching. In the image fuse step, a radiometric and color compensation stage allows the development of an automatic and seamless mosaicing system. The results show that our approach performs well, even in processing the surveillance images with low overlapping area, moving objects, shape differences and noise.