General categorization of leaves using square patches in combination with SVM classifier

The main objective of this proposed approach is to categorize various kinds of leaves despite their shapes, sizes, locations and orientations from the sub square patches in combination with SVM classifier. In spite of differentiating the one full original image from the other, initially, interest points are detected to determine the high information areas of each and every leaf image. These points are detected from the various leaf images collected from the Caltech database. Then these salient points are represented in the form of square patches which deals with variability in object shape and partial occlusions. The local representation of images that is patches are extracted over the salient points. The Ridgelet based features are computed for each and every patch. Then these features are given to SVM classifier which plays an important role in categorizing objects from backgrounds. Based on the classification results the strength of the proposed method is measured and those results show the performance of the proposed method with high recognition rate and faster processing speed.