Remote sensing image semantic segmentation method based on boundary point selection algorithm

• Main Authors: LIAO GUANGHUI, WU JIASHENG, GU ZHUJUN, WU BINGXIAO, FU QINGHUA, LIN DAIDI, ZENG MAIMAI, LIN QUANMAN, ZHAO MIN
• Format: Patent
• Language: Chinese; English
• Published: 15.11.2022
• Subjects: CALCULATING; COMPUTING; COUNTING; PHYSICS

The invention discloses a remote sensing image semantic segmentation method based on a boundary point selection algorithm, and the method comprises the following steps: firstly, carrying out the smooth filtering of an image through calculation, obtaining a smooth image, carrying out the Laplace operation of the smooth image, enhancing the image, selecting a zero cross point in a reciprocal of bivalence to carry out the edge detection criterion, and randomly selecting points; according to the method, a boundary point selection algorithm is combined with a segmentation method of edge detection and deep learning, the structure boundary points are selected on the basis of edge detection to form a boundary point diagram, boundary information of a semantic segmentation object is better reserved, the edge detection algorithm is combined, and the segmentation efficiency is improved. According to the method, the boundary of the semantic segmentation object is better matched with the real value, the problem of boundary