| Summary: | 碩士 === 國立中興大學 === 應用數學系所 === 101 === In this paper, we propose a new evolutionary algorithm which is combined with lookup table method (LUT). This algorithm can be applied in the selection for optimal multilevel thresholds of image segmentation. The objection function adopted in the algorithm is the same as Otsu’s method (between-class variance of image histogram). In our algorithm, the optimal multilevel thresholds are determined by maximizing the between-class variance. In order to increase computing speed, our algorithm is implemented in the MATLAB by using matrix operation and is combined with lookup table method to calculate between-class variance quickly.
For the verification, the numerical experiment is compared with three existing algorithm: enumerative algorithm, Liao algorithm and genetic algorithm. As the number of thresholds increase, experimental result shows that our algorithm has more advantage in comparing with other methods.
|
|---|
