基于交互式图像分割的立体匹配方法

摘要

Stereo matching technology is the key step in 3D-reconstruction system and non-contact measurement,which has a wide range of applications in computer vision. The traditional stereo matching method based on global optimization is of large computation quantity, and the framework of its algorithm is very complex. By using the method of image segmentation,it can effectively reduce the amount of computation. In order to overcome the disadvantage that traditional method cannot make use of the information of segmentation,a stereo matching method integrated human-computer interaction image segmentation is presented. Introducing the user interaction and setting the seeds,the Region Of Interest ( ROI) segmentation area is obtained by a fast graph cuts algorithm. Then the stereo matching is realized based on the network map which is constructed from the segmentation template. Compared with existing stereo matching methods, the method proposed has more accurate and less computational cost,which is verified by experiment.%立体匹配技术是三维重建系统以及非接触测量的关键步骤，在计算机视觉领域应用广泛。传统基于全局最优的立体匹配方法计算量大，算法框架复杂，通过引入图像分割等手段可以有效降低立体匹配的运算量。为了克服传统基于分割的立体匹配方法没有充分利用分割信息的缺点，提出了一种融合交互式图像分割的立体匹配方法。该方法通过引入用户交互，设置种子点并用快速图割算法完成感兴趣区域的分割提取，由分割模板建立网络图，进行立体匹配。由于仅针对分割区域构建网络图，充分利用网络图信息降低了运算量，因此，和现有的对整幅图像进行立体匹配方法相比，具有匹配准确、运算量小等优点，并通过实验进行了验证。