Face Recognition Using Gradient Texture Features

Nowadays,it is very important to recognize/identify a person for security reasons. A lot of people and organizations use biometrics for the identification of a person. Face recognition is one of the biometric identification techniques used a lot. The crucial part of the face recognition system is feature extraction and a good feature extraction technique has to be robust to noise and illumination. In this paper, we have proposed a new feature extraction technique using the texture of the image with the help of gradient features. In the proposed algorithm, at first, the face is detected using the MTCNN algorithm and the features are extracted from the detected face for the recognition purpose. Kirsch edge detection 8 kernels are applied to get edge response from the face detected and used ANOVA to find the edge similarity between each of the convoluted images of Kirsch edge response to get the final image as explained in the encoding face image section. Eigen values were obtained from the final image using Principal Component Analysis (PCA). The Chi-square method is used as a classification technique for matching the features of the test image with trained images present in the database. Here, the FERET dataset is used for training the model and testing the accuracy and we have achieved encoring results using the proposed method.