Face recognition is an interesting field for application of artificial intelligence. Three objectives can be considered in such a research study. These have to do with face localization, alignment and identification. Although becoming a mature field, many challenges still remain to be handled for face recognition. Pose and illumination variations, face similarities within a family as well as possible difficulties to find enough training samples (which also need to have high enough resolution) for each subject are some of these challenges. From obtaining the raw data to providing them to a face recognition algorithm some processes are obligatory such as locating the face in a camera image and aligning it to prevent suffering from noise due to variances of pose and illumination. In this research, different approaches for face recognition will be studied. An efficient approach is selected to design and implement in MATLAB software package. The obtained results in the research will be developed in the MATLAB and presented after analysis of the outcomes.
- CHAPTER 1: INTRODUCTION 1
- CHAPTER 2: ARTIFICIAL INTELLIGENCE STRUCTURE 3 2.1 TURING TEST 4 2.2 ARTIFICIAL INTELLIGENCE 5 2.3 MACHINE LEARNING 7 2.5 DEEP LEARNING 9
- CHAPTER 3: DIFFERENT METHODS AND APPROACHES TO FACE DETECTION 10 3.1 CLASSIFIER AND FEATURE EXTRACTION METHODS 14 3.1.1 SUPPORT VECTOR MACHINE 14 3.1.2 RESTRICTED BOLTZMANN MACHINE 17 3.1.3 DIFFERENCE-OF-GAUSSIANS FILTER 19 3.1.4 GABOR WAVELETS 21
- CHAPTER 4: THE FACE RECOGNITION SYSTEM 25 4.1 IMAGE DATASET 27 4.2 TRAINING THE SVM WITH FEATURE VECTORS 27 4.3 TESTING THE FACE RECOGNITION SYSTEM 28
- CHAPTER 5: CONCLUSION AND RECOMMENDATION 33