Automatic classification of textured images is crucial for both surface inspection problems in quality control systems and medical imaging applications. It requires sophisticated textural image features in order to distinguish between different defects or image classes. In the following paper, we report on experiments in which we automatically select adequate subsets of textural features from a large set of potential candidates. For the underlying problem of tumor cell identification, conventional selection techniques as well as genetic algorithms have been investigated on. The Gallops PVM package has been running on up to 48 machines in order to find the best feature subset.
展开▼
