Color images segmentation using the Support Vector Machines

This work introduces a new segmentation technique that is based on the Support Vector Machines (SVMs). This segmentation procedure classifies the different pixels of an image between those that belongs to a region and the others that do not belongs to it. The technique is applied to color medical images in order to separate the different color regions into them. This is a technique that requires a first time training but after this training the overall processing is performed automatically. The training is made using regions of one real image but it can be used over other images with good results. We present some examples of segmentation and discuss some problems and future work to be done based on this new technique. Key-Words: Image Segmentation, Support Vector Machines, Color Spaces, Medical Imaging, Statistical Learning, Feature Extraction