基于局部纹理特性和图像分割的分步立体匹配

A Two-step Stereo Matching Based on the Local Texture Features and Image Segmentation

为解决立体匹配难以兼顾精度和速度的问题,提出一种分步立体匹配方法。根据局部纹理特性对图像对进行灰度变换;然后利用均值漂移算法分割参考图像,以任意大小和形状的分割区域为支持窗口进行初次匹配,形成基于色彩分割的视差约束;再以固定窗口为支持窗口进行二次匹配,获得初始视差图;最后通过可信度分类优化初始视差。实验结果表明所提出的算法兼有较高的匹配速度和精度。

To solve the problem of meeting accuracy and speed requirements simultaneously, a two-step stereo matching method is proposed. According to the local texture features, a transform is adopted to the couple of images, and the reference image is segmented with mean shift algorithm, the first matching based on the support region with arbitrary shape and size was used to form a parallax constraint, then the second matching based on fixed window is adopted to obtain the initial disparity map. Finally, the disparity map is optimized by the different reliabilities classification. The experiment results show that the algorithm has higher matching speed and accuracy.