一种基于图像重投影的视图合成方法

该方法适用于摄像机绕光心旋转时虚拟视图的合成。本发明用两台摄像机在不同视点拍摄同一场景得到两幅参考图像，首先利用单应矩阵及一幅参考图像合成目标图像的一部分，然后利用三维图像变换技术，根据另一幅参考图像生成目标图像的剩余部分，得到目标图像。本发明的特点是将目标视图分为两部分来合成，其中一部分的合成最大限度地利用了单应矩阵的优良性质，提高了合成视图的质量。同时，另一部分的合成采用了三维图像变换技术生成目标图像的剩余部分，从而实现了摄像机绕光心旋转的虚拟视图的合成。-An image reprojection view synthesis method based on

This method is suitable for the synthesis of the camera optical center of rotation around the virtual view. The invention uses two cameras shooting the same scene at different viewpoints to get two reference images and the first to use a single reference should be part of the image matrix synthesis target image, and then use the three-dimensional image transformation techniques to generate a reference target image based on another image the remaining portion to obtain a target image. Feature of the present invention is divided into two parts to the target view synthesis, wherein the synthetic part to maximize use of the excellent properties homography improve the quality of the synthesized view. Meanwhile, another part of the synthesis of the remainder of the three-dimensional image using the generated transform the target image, in order to achieve the synthesis of the optical center of the camera rotates around the virtual view.