Localization Algorithm for Small Objects in an Image using Wavelet Transforms

This paper presents an algorithm for determining the coordinates of small objects appearing in a sequence of scene images based on sum and difference processing of the detailing wavelet coefficients of the current and reference images, followed by a morphological analysis of the generated pseudo-image. The effectiveness of the algorithm is compared with the difference algorithm for analyzing changes in the observation scene. It is shown that the proposed algorithm is more resistant to the effects of relative shifts of the current and reference images.