Analysis of symmetric local graph structure (SLGS) for face recognition with various thresholds

Face recognition is very useful in various fields like security, surveillance, health care, traffic guidance etc. Most of the face recognition systems do face recognition under controlled environment. The proposed system works well even in the uncontrolled environments. Here a new value is found for each pixel using a symmetric local graph structure. Histogram of the newly found pixels are determined and stored in the database. The test image is also processed in the same steps and verifies whether its histogram matches with the stored values. Finally recognition is done based on this matching procedure using the nearest neighbour classifiers like Euclidian distance, Manhattan and chi-square distance measures. Here the analysis is done with various thresholds.