提出了一种利用同一场景的两幅视差图像生成用于多视点自由立体显示的多幅视差图像的方法。首先运用SIFT匹配算法从两幅视差图像中寻找匹配点,并运用生长运算得到图像的稠密匹配;然后根据匹配点坐标对视差进行了近似计算并得到左视差图像的深度图像;最后提出一种基于投影原理的多幅视差图像生成方法,由左视差图像和其深度图像生成了应用于多视点自由立体显示的多幅视差图像,其立体显示效果良好。

An efficient method to acquire multi-view parallax images for multi-view auto-stereoscopic display from two parallax images is presented. Firstly,the feature matching of the two given parallax images by using SIFT matching algorithm is done and the dense matching the images is accomplished after the propagation of the SIFT matching points. Then,the parallaxes are approximately calculated by the coordinates of the matching points to generate the depth map of the left image. At last,a method using an original image and its depth map to acquire multi-view parallax images is proposed. Good stereoscopic images are presented on an auto-stereoscopic display device after the parallax images were synthesized.