Development of a Simulation Tool for the Face Recognition Using Feature Feedback

摘要

In this paper, we present a novel simulation tool to analyze the results from Feature Feedback-based pattern recognition method especially for the face recognition. Using Feature Feedback, we can choose an important region that affects the classification and utilize the region for the recognition. To do so, a reverse mapping is used from the feature space to the original data space. However, in Feature Feedback, there are many parameters such as number of Fisherfaces, threshold value, etc. used to obtain better results. The simulation tool described in this paper makes it easier to evaluate the effect of those parameters on the recognition rate by optionally changing their values. It also helps the students have an intuitive look on Eigenface, Fisherface or feature mask which are important for the understanding of the Feature Feedback and pattern recognition theory.