视觉显著性与知觉组织相结合的高分影像居民地提取方法  

A Residential Area Extraction Method for High Resolution Remote Sensing Imagery by Using Visual Saliency and Perceptual Organization

作　　者：陈一祥[1,2] 秦昆 张晔[3] 袁媛[1] 

机构地区：[1]南京邮电大学地理与生物信息学院,江苏南京210023 [2]安徽省智慧城市与地理国情监测重点实验室,安徽合肥230031 [3]武汉大学遥感信息工程学院,湖北武汉430079

出　　处：《测绘学报》2017年第12期1959-1968,共10页Acta Geodaetica et Cartographica Sinica

基　　金：国家自然科学基金(41501378);江苏省自然科学基金(BK20150835);安徽省智慧城市与地理国情监测重点实验室开放性课题基金资助课题(2016-K-03Z);南京邮电大学引进人才科研启动基金(NY214196)~~

摘　　要：受人类视觉认知机制的启发,提出了一种利用视觉显著性与知觉组织相结合的高分辨率遥感影像居民地提取方法。首先利用认知物理学中的数据场构建居民地的视觉显著性模型,并通过自适应阈值法实现候选居民地的自动提取,然后利用多尺度小波变换的高频特征实现居民地的知觉组织,最后通过集合交运算提取同时满足这两种视觉机制的居民地。通过ZY-3和Quickbird两种高分传感器的影像数据集进行居民地提取试验,验证了该方法的有效性。Inspired by human visual cognitive mechanism,a method of residential area extraction from highresolution remote sensing images was proposed based on visual saliency and perceptual organization.Firstly,the data field theory of cognitive physics was introduced to model the visual saliency and the candidate residential areas were produced by adaptive thresholding.Then,the exact residential areas were obtained and refined by perceptual organization based on the high-frequency features of multi-scale wavelet transform.Finally,the validity of the proposed method was verified by experiments conducted on ZY-3 and Quickbird image data sets.

关 键 词：高分辨率遥感影像 视觉显著性 数据场 知觉组织 小波变换 居民地提取 

分 类 号：P237[天文地球—摄影测量与遥感]