基于PCA-PSO的图像匹配和拼接
Images Matching and Stitching Based on PCA-PSO
-
摘要: 针对传统图像匹配算法和基于PCA的图像匹配算法误匹配较高进行改进.首先, 利用SIFT生成128维描述子向量矩阵, PCA即主分量分析法对矩阵进行降维.然后, 以一幅图的每个特征点描述子向量为基准, 在另一幅图的特征点描述子矩阵中利用PSO即粒子群优化算法寻找基准图特征点的全局最优解即匹配对.最后利用这些匹配对进行图像拼接.与基于距离的传统匹配法相比, 本算法匹配正确率更高, 图像拼接质量更好.Abstract: Traditional images matching usually got some wrong matchers, which mean need to improve.First of all, get the 128-D descriptors according to SIFT, Make the dimensions less by using PCA and then find the best matchers by using PSO and finally start to stitch. Contrast to the traditional matching method and the images matching method based on PCA, the algorithm in this paper proved to be more accurate and it makes the following stitching better as well.
下载:
