The paper proposes a new method that combines the self-adaptation mechanism and Adaptive Particle Swarm Optimization (APSO) algorithm for binary image registration, taking into consideration the problem of a particular component of banking security which involves the client's signature. The main aim is to register two binary images, considering the affine perturbation model that consists of a mixture of shear and rigid deformation. In most of the cases the image to be recognized is different from the stored image, from geometrical point of view. A self-adaptive APSO method was developed to align the input image to the target one. The light intensity of an individual is defined in terms of mutual information that is calculated between the transformed image and the target image. The proposed registration methodology is evaluated using qualitative and quantitative measures. In the final part of the paper, the experimental results are reported together with some concluding remarks regarding the quality of the resulted methods.