On the Robustness of Recognition of Degraded Line Images

Pattern recognition algorithms are not robust compared to human vision. Our recognition systems seem to work like some impaired human cortex, showing agnosia defects. Robustness of human vision is probably the consequence of interaction and feedback of the multiple regions of the vision cortex. Recent results on binary images of handwritten cheques show the interest of approaches based upon multiple operator combinations and soft decisions, recalling the working of a human vision cortex.