Color invariant object recognition using entropic graphs

We present an object recognition approach using higher order color invariant features with an entropy based similarity measure. Two methods to compare images are: 1) histogram matching and 2) assuming a fixed probability density function. Histogram bin size is usually set in an ad-hoc manner, where the best bin size for a specific application is experimentally determined. Another solution to image matching is found in assuming prior knowledge about the probability distributions. However, not all processes can be described with a fixed parameterized model. Furthermore, assuming one distribution might severely over-simplify the complexity of the data. Entropic graphs offer an unparameterized alternative. An entropic graph estimates entropy and related information theoretic measures from a graph structure. We extract color features from object images and make them invariant to shadow and shading. We employ entropic graphs for probability density estimation. The Henze-Penrose similarity measure is used to compute the similarity of two images. We evaluate our method on the ALOI collection, a large collection of object images. This object image collection consists of 1,000 objects recorded under various imaging circumstances.