Object recognition with the hybrid evolutionary algorithm and response analysis in security applications

The recognition of an object in a scene is a common and important task of electronic imaging arising in many defense and security applications. When the image of the sought object is significantly distorted, and the image of the scene is cluttered, noisy, and contains many objects, the commonly used methods based on correlation and comparison of the feature vectors of the images can show poor performance. The approach utilizing the particular model of the hybrid evolutionary algorithm based on image response analysis is proposed to solve the object recognition problem formulated as the global optimization problem. The computational experiments with two-dimensional grayscale images show that the proposed approach can solve complex object recognition problems. It is able to discriminate between objects having a high degree of similarity, and to detect the sought object in the large cluttered and multi-object scene.