Th.E. Schouten

Modeling human color categorization

A unique color space segmentation method is introduced. It is founded on features of human cognition, where 11 color categories are used in processing color. In two experiments, human subjects were asked to categorize color stimuli into these 11 color categories, which resulted in markers for a Color LookUp Table (CLUT). These CLUT markers are projected on two 2D projections of the HSI color space. By applying the newly developed Fast Exact Euclidean Distance (FEED) transform on the projections, a complete and efficient segmentation of color space is achieved. With that, a human-based color space segmentation is generated, which is invariant for intensity changes. Moreover, the efficiency of the procedure facilitates the generation of adaptable, application-centered, color quantization schemes. It is shown to work excellently for color analysis, texture analysis, and for Color-Based Image Retrieval purposes.

Fast exact Euclidean distance (FEED) transformation

Fast exact Euclidean distance (FEED) transformation is introduced, starting from the inverse of the distance transformation. The prohibitive computational cost of a naive implementation of traditional Euclidean distance transformation is tackled by three operations: restriction of both the number of object pixels and the number of background pixels taken in consideration and pre-computation of the Euclidean distance. Compared to the Shih and Liu 4-scan method the FEED algorithm is often faster and is less memory consuming.

FIELDNET, A Dynamic Network For Pattern Classification

A new kind of neural network (FIELDNET) is introduced. It is trained with supervised data and grows its single hidden layer (also called its codebook) during learning. Each hidden neuron belongs to an output class and has a kind of force field around it, which determines how well an input pattern belongs to its class. Only one learning parameter is present. The performance of FIELDNET is compared with three other neural networks on three real-world data sets. It will be shown that FIELDNET is capable of very fast learning while achieving high classification rates.