关键词: 目标提取, 车辆边缘, 遗传算法, 航摄影像, 模糊分割

Abstract: Nowadays, vehicle object extraction of remote sensing image and its potential applications become a hot topic. However, the optical information of vehicles is seriously weakened by a series of factors, such as environment and the velocity of the vehicles, so that the extraction accuracy of vehicles is accordingly reduced. To this end,a vehicle extraction method of remote sensing image using genetic algorithm and mathematical morphological operation has been proposed. First, a pre-clustering operation of remote sensing image is done by using histogram-based technique, and optimizing the initial cluster centers applied the genetic algorithm to improve the accuracy of segmentation results. Then, the mathematical morphological operation is used to extract the edges of vehicles. The aerial images are tested to extract vehicles using the proposed approach. The experiment results show that this method can identify the situation of vehicles on the ground as well as extract the contour line accurately.

Key words: object extraction, vehicle extraction, genetic algorithm, aerial images, fuzzy segmentation