Salient Segmentation based Object Detection and Recognition using Hybrid Genetic Transform

摘要

Object detection and recognition is an effective and fundamental technique used to track the objects accurately in complex scenes. Over the last decade, object analysis has caught the attention of researchers to explore and cover the aspects of object detection and recognition related problems in the technologies such as robotics, surveillance, agriculture, medical and marketing. In this paper, we present a unique method for accurate object recognition. Firstly, the clustering of similar colors and regions is achieved by applying K-mean clustering algorithm. Secondly, segmentation is performed by merging the previously achieved clusters, which are similar and connected. Thirdly, Generalized Hough transform is used for the detection of salient objects. Finally, Genetic algorithm is applied as recognizer engine to recognize the salient objects under different environmental settings. The accuracy of our experimental work has been evaluated on the benchmark dataset MSRC.