Discriminant simplex analysis

Image representation and distance metric are both significant for learning-based visual classification. This paper presents the concept of κ-Nearest-Neighbor Simplex (κNNS), which is a simplex with the vertices as the κ nearest neighbors of a certain point. κNNS contributes to the image classification problem in two aspects. First, a novel distance metric between a point to its κNNS within a certain class is provided for general classification problem. Second, we develop a new subspace learning algorithm, called Discriminant Simplex Analysis (DSA), to pursue effective feature representation for image classification. In DSA, the within-locality and between-locality are both modeled by κNNS distance, which provides a more accurate and robust measurement of the probability of a point belonging to a certain class. Experiments on real-world image classification demonstrate the effectiveness of both DSA as well as κNNS based classification approach.