Boundary identification and characterization through density differencing

摘要: Methods and systems are provided for identifying characterizing boundaries in digital images other data sets. At least two points within a image selected. Clusters of sample around the chosen. A pixel value histogram is generated each point cluster. These histograms define frequency functions. The distance between functions, called “density distance” calculated analyzed. nature density then used to determine if areas relatively uniform, or they indicate presence boundary. To further refine procedure, method may be applied iteratively by modifying cluster variances, sizes, distributions, reanalyzing functions generated.