基于非下采样Contourlet变换的交通图像融合方法研究

电视监控是智能交通监控系统的一个重要组成部分，其目的是通过对视频交通图像处理进行车辆、行人、环境等的监控. 根据非下采样Contourlet变换具有多分辨率、多方向性和平移不变性的特点，提出了一种基于非下采样Contourlet变换的多聚焦交通图像融合方法. 融合策略采用低频系数取平均，高频系数基于邻域、兄弟和父节点信息的区域特征衡量取最大值法. 将本文的方法与小波变换、脊波变换及Contourlet变换相比较，实验结果表明，该方法取得了更好的融合效果，提高了图像质量，满足智能交通监控系统的要求.

Traffic Image Fusion Algorithm Based on Non-Sampled Contourlet Transform

The TV monitor is an important part of the intelligent transportation monitoring system. Its purpose is monitoring the vehicles, pedestrians, and environment through the video traffic image processing. According to the non-sampled Contourlet transform with the characteristics of multi-resolution, multi-directional, and translational invariance, the paper proposes a multi-focus image fusion algorithm based on the non-sampled Contourlet transform. The fusion strategy is that the low-frequency coefficient is used to calculate the average and the high-frequency coefficients are taken the greatest regional characteristics weight based on neighborhood, brother and parent information. After comparing the present method with the wavelet transform, the ridgelet transform, and Contourlet transform, the experimental results prove that the method mentioned in this paper can achieve a better fusion of results and can improve the image quality and meet the requirements of intelligent transportation control system.