A Novel Image Classification Using Orthogonal Feature Distribution Matrix

The task of image classification has been studied in many ways using variety of features like shape, color, geometric, robust and etc. but the classification process uses only the features contained in more volume in a region but missing the features distributed throughout the image. We propose a new approach for image classification, using orthogonal feature distribution matrix. The orthogonal feature distribution matrix represents set of feature points around different axes of the image spectrum where there is ineligible volume of feature presents. The proposed method splits the axes into sixteen and at each axes, feature points are identified and the features of those points are converted into orthogonal feature distribution matrix. The OFDM (Orthogonal Feature Distribution Matrix) simplifies the process of classification, where it could be executed with least processing time and avoids missing values. In earlier approaches there are situations where the algorithm misses sheared features on classification, where as in our approach the features which gets sheared also will be obtained to produce more efficient classification. The proposed method has produced efficient classification both on time and space complexity with higher rate of classification accuracy.