Underwater sonar image multipath effect noise removal method based on geometric probability

The invention relates to the technical field of sonar image noise removal, and particularly discloses an underwater sonar image multipath effect noise removal method based on geometric probability, and the method comprises the steps: carrying out the preprocessing and denoising of a sonar image, carrying out the clustering processing of the echo points of the image after preprocessing and denoising, and carrying out the clustering processing of the echo points of the image; and removing multi-path echo effect points from the clustered image. When the method is used for detecting and removing the multipath effect points on the single two-dimensional sonar picture, the accuracy of entity pixels is kept, the effect of greatly reducing the multipath effect is achieved, and the obvious noise reduction effect is achieved; the image processed by the method is used for three-dimensional reconstruction, and the result is greatly improved. 本发明涉及声呐图像噪声去除技术领域，具体公开了一种基于几何概率的水下声纳图像多路径效应噪声去除方法，步骤包括：对声纳图像进行预处理去噪，对预处理去噪后的图像回波