A Novel Edge Detection Method Based on PCA

By analyzing the direction characteristic of principal component analysis (PCA), we propose an edge detection method based on PCA. Using Karhunen–Loëve transform, PCA transforms the original dataset into lower-dimensional feature data. The transform has directivity both on energy accumulation and data selection. The author points out and proves the two direction characteristics. In this paper, authors proposed a method of edge detection based on PCA. Authors used twice PCA operations, called TPCA, first by process the image by PCA and second by process the transpose of the image by PCA. Experimental results are shown by using some typical images.