Active Contour Model with Splitting Characteristics for Multiple Area Extractions and its Hardware Realization

In this paper, a new sampled active contour model (ACM) and its hardware realization method are proposed. The proposed Sampled-ACM is a virtual closed curve with some contour points to extract a specified area in images. Forces, such as "pressure", "attraction", "repulsion" and "vibration factor" work on every contour point of the proposed Sampled-ACM. The proposed Sampled-ACM only accesses the memories on which the contour points locate, the number of the memory access is reduced, and the faster processing speed is performed than ordinal contour models such as Snakes. The proposed Sampled-ACM has characteristics which reverses direction of attraction by the basis of specific conditions and also has splitting characteristics for multiple area extractions. The proposed Sampled-ACM is realized on FPGA (field programmable gate array) to construct a real-time hardware equipment. The experimental results to confirm the fast processing abilities of the proposed Sampled-ACM are also included