针对由于光照变化造成的立体图像误匹配问题,提出一种使用同态滤波和census非参数变换相结合的立体匹配算法。首先,使用同态滤波对极线矫正后的立体图像对进行预处理,减少光照变化对图像的影响。然后,在稳态匹配概率的基础上结合census非参数变换构造代价函数,计算初始视差图,通过左右一致性检测,滤除误匹配点。最后利用边缘感知视差传播和测地距滤波对不一致的视差点进行填充,得到精确度较高的视差图。实验数据表明,该算法减少了光照对视差计算结果的影响,提高了计算的准确性。
Aiming at the mismatch problem caused by illumination variation,a stereo matching algorithm based on homomorphic filtering and census nonparametric transform is proposed. Firstly,homomorphic filtering is used to preprocess the stereo image,reducing the influence of illumination changes on the image.Then,based on the steady-state matching probability( SSMP) and the census nonparametric transformation,the cost function and the initial disparity map are constructed. The inconsistent pixels with the leftright consistency check are filed out. Finally,EDP and geodesic filter are used to fill into the disparity map with inconsistent pixels,and the disparity map with higher accuracy is obtained. The experimental results show that the proposed method has simple mechanism,high computational efficiency,thus reducing the influence of illumination variation to some extent.