AUTOMATIC ADAPTATION OF IMAGE SEGMENTATION CONTROL PARAMETERS FOR OUTDOOR SCENES

This paper proposes a method for automatic adaptation of segmentation control parameters based on Genetic Algorithms. The goals of automatic adaptation of segmentation parameters in this research are to provide continuous adaptation to normal environmental variation conditions such as (time of day, weather) to exhibit learning capability and to provide robust performance when interacting with a dynamic environment. The research intended to adapt the performance of a well known existing segmentation method (Recursive region growing Algorithms) across a wide variety of environmental conditions which cause changes in the image characteristics and maximizes the segmentation quality measures. Experimental results are given that indicate the ability to adapt the segmentation performance in outdoor color image with high quality segmentation result in a minimal number of generations.