Facial Expression Classifier Using Better Technique: FisherFace Algorithm
Nishanth S. Hegde, S Preetha, S Dhiraj Bhagwat · 2018
Facial recognition is one of the most secure modes of authorization which uses samples of photographs or videos to authorize a person's identity. Facial recognition has its own weaknesses owing to various criteria like difference in number of image samples, clarity, brightness of the photograph etc. Hence a lower accuracy rate compared to other modes of authentication like Fingerprint Recognition, which has its own uniqueness and greater performance. This paper selects Fisherface algorithm which provides the best performance, greater accuracy, and use it to verify humans by comparing the facial recognition result with our database. The two important characteristics, which are SFMR (Successful Face Match Rate) and USFMR (UnSuccessful Face Match Rate) to analyze the performance of the algorithms for different block sizes using Gaussian Adaptive Threshold method. Facial Recognition algorithms is based on a number of tests with varying conditions like environment, lighting, thresholds, line of sight etc. The FisherFace algorithm is used to detect different human emotions or expressions and the accuracy with which the emotions are recognized and calculated.