This paper discusses the problem of introducing feedback in multilevel machine-vision systems. Based on a theoretical-informational analysis, it is shown that feedback is needed in such systems because the individual components of the quality criterion common to the entire system are optimized at different levels. As a result, the decisions made at the earlier stages of the analysis use only part of the components of this criterion, and this makes these decisions less than optimal. An approach is proposed to introduce feedback as a method of achieving the global optimum of the informational quality criterion by iterative improvement of the decision. Based on these results, it is shown that the information contained in the contours is inadequate for a robust construction of the structural elements.
© 2007 Optical Society of AmericaPDF Article