Image features make maxima in parameter spaces. Algorithms {hill-climbing methods} can find local maxima that exceed threshold amount. If maximum is at feature parameter-space location and exceeds threshold, algorithm states that feature is in image and identifies location. Hill-climbing methods can become stuck at local maxima and fail to find more-important maxima.
Hill-climbing algorithms {Broyden-Fletcher-Goldfarb-Shanno method} (BFGS method) can improve quasi-Newton method.
Hill-climbing algorithms {Newton's optimization method} {Newton optimization method} can solve unconstrained non-linear optimization problems using function second-order partial-derivative Hessian square matrices, to find local maxima at locations where derivatives become zero.
Hill-climbing algorithms {quasi-Newton method} can simplify Newton's optimization method by simplifying Hessian matrix.
Hill-climbing algorithms {watershed algorithm} can find maximum gradient from center pixel to eight surrounding pixels and move to pixel with maximum gradient. If new pixel is minimum or is lower than threshold, stop and assign original pixel to same group as second pixel. If new pixel is not minimal and is not lower than threshold, find maximum gradient from second pixel to eight surrounding pixels and move to pixel with maximum gradient.
3-Computer Science-Systems-Computer Vision-Algorithms
Outline of Knowledge Database Home Page
Description of Outline of Knowledge Database
Date Modified: 2022.0225