Self-tuning Parameters in Boundary Contour System (BCS)

P. Martínez, M. Pinzolas, J. López-Coronado (Spain)


Computer Vision, Cognitive Vision, Neurobiological Systems.


In this paper, a procedure for auto adjusting critical parameters in the perceptual image pre-processing algorithm called Boundary Contour System / Feature Contour System (BCS/FCS) is presented. This task can not be hand-made in an efficient manner, because of the great number of parameters involved. In this work, a second-order optimization method called Levenberg Marquardt has been used. The Levenberg-Marquardt algorithm can be thought of as a trust-region modification of the Gauss-Newton algorithm, and it needs an error index to quantify the algorithm performance. With this purpose, a set of prototypical images have been created. The accurate tuning of the values of the BCS/FCS parameters produces an enhancement of the results onto the images treated, as shown in this paper. .

Important Links:

Go Back