论文部分内容阅读
针对高光谱图像的非监督目标检测问题,提出了一种基于混沌粒子群优化(PSO)投影寻踪(PP)的检测方法。混沌PSO可加快PP过程,得到更精确的最佳投影方向。利用自适应波段选择方法进行高光谱图像降维。依据对异常分布敏感的偏度和峰度设计投影指标,并采用混沌PSO搜索最佳投影方向,由此可有效地将目标信息投影至低维空间。采用直方图分割的方法从投影图像中提取出目标。针对大量图像进行了实验及检测效果的定性与定量评价,并与遗传算法PP法、RX方法的检测结果作了比较。结果表明,能够更有效地检测出高光谱图像中的目标,且所需运行时间大为减少。
Aiming at the problem of unsupervised detection of hyperspectral images, a detection method based on chaotic particle swarm optimization (PSO) projection pursuit (PP) is proposed. Chaotic PSO can speed up the PP process, get more accurate best projection direction. Using adaptive band selection method for hyperspectral image dimensionality reduction. According to the skewness and kurtosis which are sensitive to the anomaly distribution, the projection index is designed and chaotic PSO is used to search the best projection direction, which can effectively project the target information into the low-dimensional space. The method of histogram segmentation is used to extract the target from the projection image. A large number of images were tested and the results of qualitative and quantitative evaluation were evaluated. The results were compared with the results of PP and RX methods. The results show that the target in the hyperspectral image can be detected more efficiently and the required runtime is greatly reduced.