An Adaptable Threshold Decision Method

Otsupsilas thresholding method (OTM) is one of the most commonly used thresholding methods. Unfortunately, the threshold obtained by OTM is biased in favor of the class, whose standard deviation or quantity of data is larger. Besides, one may adopt distinct thresholds in different applications for a same data set. Accordingly, this paper proposes an adaptable threshold decision method (ATDM) to provide the most appropriate thresholds for assorted applications. This paper also proposes a PSO (particle swarm optimization) based parameter detector (PBPD) to decide the fittest parameters which are used by ATDM. Image segmentation extracts the regions of interest from an image for follow-up analyses, and thresholding is one important technique for image segmentation. This paper will employ ATDM to detect the object contours in an image in order to investigate the performance of ATDM. The experiments show that ATDM can give impressive segmentation results.