Analysis of selecting the optimal threshold in image segmentation based on the evolution of feature field
Shuqin Liu, Jinye Peng
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12B) 465-468
Information Science and TechnologyCollege of Northwestern University, 710127, China
In order to select the optimal threshold in image segmentation, this paper raised an image segmentation method based on the data field evolution mechanism. Integrating the local gray feature with the metric texture, it enabled to extract sufficient image information. Imaging that every pixel with multi-features was a particle with physical meaning, it built a feature field in the space of image feature. Under the supposition that the optimal threshold was the potential direction of evolution, particles would self-adapt to attract or repel each other because of the interaction in the dynamic data field among particles. In this way, the co-evolution was achieved and we further got the segmentation result. The experiment result shows that this method acquires quite good segmentation performance and is quite practical, without significantly increasing the time complexity.