Isolated forest underwater artificial target detection method based on genetic algorithm

The invention discloses an isolated forest underwater artificial target detection method based on a genetic algorithm. The method comprises: firstly, reading an original sonar image file collected byan underwater robot; secondly, normalizing the original sonar image data and filling missing values; then, defining a population size and iteration times, initializing each individual of the population in the genetic algorithm, extracting sample data, and calculating abnormal data points and normal data points in the image data by establishing a separation tree and an isolated forest; and finally,performing training iteration through a population in the genetic algorithm, and if the number of iterations reaches a specified number of iterations, outputting an optimal solution; and if not, returning to continue the steps. According to the method disclosed by the invention, self-adaptive parameter setting is provided, so that precision loss caused by manual setting is avoided, and compared with an original isolated f