摘要
图像Morphing是图像之间插补变换的一类被广泛使用的技术 ,这种方法是基于两幅图像之间的像素位置和颜色的内插 ,但对 3D场景或物体 ,它不能保证所生成图像的真实性 ,即不满足透视性质 ,在此介绍一种图像插补变换的新方法 :视图Morphing技术 ,它可以保证生成图像的真实性 ,且不需要知道图像的 3D信息 ,这避免了图像的 3D重建 ,因而具有极大的应用价值。一般的视图Morphing技术需要知道图像的投影矩阵 ,可是用目前的技术 ,从图像直接获得投影矩阵是很困难的。视图Morphing技术事先不需要知道图像的投影矩阵 ,就可以对摄像机刚体运动所获得的两幅图像进行插补运算 ,获得中间视点的图像。
Image morphing is widely adopted in image interpolating. The method is basing on the interpolation of the position and the color of pixels of two images. However, in the case of the scene or object of three dimensions, the reality of the inborn image can't be ensured, that is, the perspective properties can't be satisfied. In this paper, we introduce a new method of image interpolating, View Morphing. This method ensures the reality of the inborn image and the three dimensions information of images are not needed. This method avoids the reconstruction of three dimensions objects, so it has great value. General view morphing technology needs to know the projective matrices of images,however,it is very difficult to acquire the matrices directly from images using present technologies. The view morphing technology in this paper enables us to obtain the image of the middle view point by interpolating the two images which are obtained by a camera's rigid motion without the knowledge of the projective matrices.
出处
《微机发展》
2002年第5期44-46,共3页
Microcomputer Development