In this paper, a new edge detector (boundary extractor) is proposed based on finding major change points in a local one-dimensional window of the image intensity values of the rows or columns. The approach amounts to separating the pixels in the window into sets or regions of constant intensities with the edge pixels providing transition points. The edge points are found based on partitioning the interval in an optimal way using dynamic programming with an appropriate cost function. Different cost functions are introduced for the algorithm with simulation results that show the detector's effectiveness even in the presence of noise

Many signal processing problems can be solved by maximizing the fitness of a segmented model over all possible partitions of the data interval. This letter describes a simple but powerful algorithm that searches the exponentially large space of partitions of N data points in time O(N/sup 2/). The algorithm is guaranteed to find the exact global optimum, automatically determines the model order (the number of segments), has a convenient real-time mode, can be extended to higher dimensional data spaces, and solves a surprising variety of problems in signal detection and characterization, density estimation, cluster analysis, and classification.

Most existing deconvolution techniques are incapable of determining phase properties of wavelets from time series data; to assure a unique solution, {\em minimum phase} is usually assumed. It is demonstrated, for moving average processes of order one, that deconvolution filtering using the absolute value norm provides an estimate of the wavelet shape that has the correct phase character when the random driving process is nonnormal. Numerical tests show that this result probably applies to more general processes.