Model-Based Recognition Method for Circular Boundaries

To recognize circular boundaries, a mixed probability density function that can generate the circular spatial point pattern is built up. In terms of the above, the probability density function, a circular boundaries recognition method is constructed, It consists of three parts: viz.: denoising; estimating the parameters of mixed probability density function, and identifying the number of clusters via BIC. The new method overcomes the limitation of fuzzy C-shell clustering. simulation study provided some promising results.