基于结构光的多投影显示系统图像对准算法

Image Alignment Algorithm for Multi-projector Display System Based on Structured Light

为了解决多投影显示系统的图像对准问题,提出了一种基于结构光投影技术的对准算法.该算法通过求取等相位线的交点来建立投影机图像与相机拍摄图像间的映射关系,对于投影机图像上任一像素点,有水平和垂直2个相位值.在相机图像的水平相位展开图中找出与该点水平相位值相等的离散坐标点的位置,拟合出一条水平等相位线,同理,可以得到它的一条垂直等相位线;求出水平和垂直等相位线的交点,即为该点对应于相机图像中的像素点.该算法无需知道显示墙的解析表达式和相机的内部参数,适用于任意光滑的曲面显示墙.对准后图像与原始图像的对比实验表明:峰值信噪比达到27.904,结构相似度为0.972 4,说明该算法能有效提高图像的对准精度.

In order to solve the problem of image alignment in a multi-projector display system,an image alignment algorithm based on the structured light projection technique was proposed.The algorithm uses intersection points of phase-equal lines to establish the mapping relation between projector images and camera images.For an arbitrary pixel point on projector images,it has a horizontal phase value and a vertical phase value.In horizontal phase unwrapping diagram of camera images,a series of discrete coordinate points with a phase value equal to the horizontal phase value of this pixel point is found,and these discrete points are fit to a horizontal phase-equal line.In the same way,a vertical phase-equal line of the pixel point can be obtained.And the intersection point of the horizontal and vertical phase-equal lines is the pixel point on camera images corresponding to the projector pixel.The analytic expressions of display wall and the internal parameters of a camera are not necessary for the proposed algorithm,so the algorithm is suitable for curved display walls with arbitrary smooth surface.The comparison experiments between aligned images and original images show that the peak signal to noise ratio reaches 27.904 and the structure similarity is 0.972 4 to demonstrate that the proposed algorithm can improve image alignment accuracy effectively.