Abstract

Image retrieval from image databases is usually performed by using global image characteristics such as texture or colour. The use of local image information is highly desirable when only part of the image is of interest, but global approaches are not well suited to this. An original solution was introduced in [11] using invariant local signal characteristics. This paper extends this contribution by extending the set of invariants considered to allow illumination change. Then it is shown that the invariant distribution is far from uniform and a probabilistic indexing scheme is proposed. Experimental results validate the approch and the different method are discussed. The main result is that it is much more valuable to increase the discrimant power of the vector used to perform the indexing process; The Bayesian decision improves the standard method, but this improvement is much more limited than expected.

Citations