Research On Single-foot Footprint Principal Axis Align

This paper makes use of Principal Components Analysis (PCA) to research the hard ground flattie single-foot footprint principal axis align .PCA is multidimension orthogonal linear transformation which based on statistical characterization. In practical application, PCA is called K-L transform. To discrete image K-L transform is called Hotelling transform. Generally, PCA is used to compress data. In this paper,it is used to align single-foot footprint principal axisIn order to segment image ,there have two designed method which one is a fixed threshold value algorithm and the other is an algorithm based on the Otsu .Then the opening and closing algorithm of morphology are assembled to fill the footprint region. Experimental results demonstrate that these designed image preprocessing algorithm fit with the subsequent algorithm's require basically for the footprint sample with large noise and the small noise .To align the footprint length of the footprint region on the horizontal axis or the vertical axis, the footprint centroid and eigenvector is calculated with the 2-D Principal Components Analysis(PCA) , and then the Hotelling transform is used. First conclusion in these experiments is that the eigen axis is exclusive for the facultative form, and the eigen axis could describe the principal axis in substance for the footprints with integrated boundary. Second conclusion is that the centroid and eigenvector of the region which boundary is integrated but the heel is missing have some change. Third conclusion is that the male hard bottom shoes with heel after image preprocessing is cut in middle and it's centroid and eigenvector have angle error and centroid error. Forth conclusion is that the travel shoes and casual shoe's footprint have missing region in foot bow after image preprocessing, it's eigen axis could describe the footprint's chief axis, but it's centroid have error. The last conclusion is whatever the footprint with integrated boundary,the footprint which middle is cut and the footprint which foot bow is missing, the centroid and eigenvector could simply describe the footprint's some characters, so these may describe the footprint character as the footprint's eigenvalue.