A Genetic Algorithm for the Identification and Segmentation of Known Motion-Blurred Objects

This paper presents a Genetic Algorithm based technique capable of identifying moving objects whose image is blurred due to fast relative motion with respect to the acquisition camera. Morever, also teh extension of the motion and the rotation of the object during the acquisition time can be accurately inferred. The proposed method is applicable when then geometric properties of the object were previously recorded in a database. Extensive testing shows that the proposed algorithm yields high success rates of correct identification of both the bottle species and of its motion with a limited number of chromosomes. The computing time is reasonably fast and the algorithm can be applied in real-time applications.