| Summary: | 碩士 === 義守大學 === 電子工程學系 === 88 === The Genetic algorithm mimics the process of natural evolution, which the driving process for the emergence of complex and well-adapted organic structures. In the natural world, after computing with each other the fittest individuals survive and reproduce next generations. Genetic algorithms can search the optimal and stable solutions of the complex problems in diverse fields as the fittest individuals parallelly.
In this thesis, the method for auto-selection of the features of the ultrasonic images is proposed. The algorithm can be divided into the three steps: At first, features were of the original image including the texture features and the statistical features extracted by convolution of Laws’ Feature masks and calculation of the moments respectively. Then, main features were selected by Genetic Algorithm (using cost function). Finally, different tissues were classified by K-means clustering or Self-organizing feature maps.
The ultrasonic images were categorically segmented into several parts by the auto-classification with Genetic Algorithm. This auto-feature-selection system based on the genetic algorithm can improve the identification of ultrasonic images, therefore can assist the diagnose by the ultrasonic images.
|
|---|
