一种图像匹配中SSD和NCC算法的改进

论文针对图像匹配计算中的SSD和NCC算法提出一种基于递推增量计算的改进算法。分析和利用计算图像每个像素的SSD和NCC值时的模板在水平方向和竖直方向上具有的平移特性,以及前后上下像素模板的相互关系,利用已计算的值,来计算新的像素点的SSD和NCC值。定量分析和仿真结果显示,该算法能够有效地降低计算量。另外由于只是从计算方法上对这两种算法进行了改进,并未改变其原理,所以该方法能够保证SSD和NCC算法在立体匹配计算时本身的特性没有改变。

A Computational Optimization Method of the SSD and NCC Algorithms of the Stereo Matching

This paper proposes computational optimization method based on the incremental computation of the SSD and NCC algorithms for the stereo matching.This paper analyzes and utilizes template translation when computing the SSD and NCC of every pixel on the horizontal and vertical orientation,and the relation of the pixel template with its neighbors pixels templates.By means of the computed value.The new pixels' values are computed rather than computing each values completely from scratch.According to the quantitative analysis and the simulation results,our computational optimization of algorithms reduces the computing complexity efficiently.Because we just improve the computing method of the SSD and NCC but not change their principles,so the matching accuracy of the algorithm still can be assured.