In order to classify the spoofed as well as non-spoofed faces from images, the face spoof detection technique has been proposed. In order to analyze the textual features present within a test image, the DWT algorithm is applied. Due to the effectiveness, non-parametric as well as ease of implementation, there are numerous applications that have been utilizing this method. However, the time that is utilized for classification is longer and the optimal value of k is difficult to be calculated. On the basis of data, the value of k is generated here. The effect of noise present on classification is minimized by larger values of k. However, there is less distinct boundary generated amongst the classes. The traditional KNN mechanism is enhanced by using several K-values of various classes in order to overcome such drawbacks. The WKNN mechanism is utilized in order to enhance the performance of KNN. In order to analyze the proposed approach, comparisons are made amongst the proposed and existing mechanisms in terms of accuracy and execution time.
Face spoof, KNN, SVM