Feature extraction with common vector approach

The purpose of this study is to analyze and to comment selected feature region ? recognition rate, null subspace of all feature space ? recognition rate and pixel usage ? recognition rate relationships by doing feature extraction with linear subspace reduction techniques such as common vector and discriminative common vector approaches. For this, the regions representing the classes in the best way were determined with a mathematical approach, a face recognition application extracting face features and applying the common vector and discriminative common vector approaches was developed. Face recognition application was experimented with various databases such as AR, Yale B, ORL, Faces94, Faces95 ve Faces96. In the study, the purposes were realized by implementing above techniques, efficient simplifications were succeeded at pixel usage rates with selected pixels beginning at the middle regions of eyes in answer to acceptable losses in recognition rate. This touch enables faster implementation, less storage area and suitability to realize for discriminative common vector approach. In conclusion, data gotten from application were evaluated and commented by being combined after each experiment.Keywords: Common Vector Approach, Discriminative Common Vector Approach, Feature Selection, Feature Extraction, Face Recognition, CVA, DCVA