Many colored object recognition methods tend to fail when the incident illumination varies. In the context of image indexing, a method is presented, which does not depend on lighting conditions. A new approach for indexing images of persons moving in areas in where the acquisition is monitored by color cameras is developed to cope with the variations of the lighting conditions. We consider that illumination changes can be described using a simple linear transform. For comparing two images, we transform the target one according to the query one by means of an original color histogram specification based on color invariant evaluation. For the purpose of indexing, we evaluate invariant color signatures of the query image and the transformed target image, through the use of the color co-occurrence matrices. Results of tests on real images are very encouraging, with substantially better performance than those of other methods tested.
展开▼
