Spatial domain entropy based local feature extraction scheme for face recognition

In this paper, an entropy based feature extraction algorithm for face recognition is proposed, which efficiently exploits the local spatial variations in a face image. Instead of considering the entire face image as a whole, first considering small segments of image along vertical direction, the entropy values are measured for each segment and thereby a vertical signature is obtained. It is shown that the variation in face geometry along the vertical direction can be closely mapped with the extracted vertical signature. In a similar fashion a horizontal signature is obtained. Both the horizontal and vertical features are employed together in the proposed feature space, which provides a very high within class compactness and between class separation. Extensive experimentations have been carried out upon standard face databases and the recognition performance is compared with some of the existing face recognition schemes. It is found that the proposed method offers not only computational savings but also a very high degree of recognition accuracy.