论文部分内容阅读
遥感图像会因成像系统和地理环境而产生噪音,这些噪音将会影响从TM图像中提取森林信息的精确性和有效性。消除噪音对图像的分类十分重要。本研究的目的是评估应用Landsat 5 TM图像提取森林相关信息时,目前所使用的空间滤波处理方法的有效性。对低通滤波、中值滤波、均值滤波、求和滤波、增强型自适应滤波五种空间滤波方法做以检验。通过设计一系列的评估指数,分析每种噪音消除方法的平滑能力、边界保持和增强能力。基于所设计的评价指数和图片对比表明,对林区土地利用和森林类型分类而言,求和滤波(D=1)和增强型自适应滤波是消除Landsat 5 TM图像噪音的最有效的方法。图1表2参29。
Remote sensing images can be noisy due to the imaging system and geography that will affect the accuracy and validity of extracting forest information from TM images. Eliminating noise is very important for the classification of images. The purpose of this study is to evaluate the effectiveness of the spatial filtering methods currently used when extracting forest-related information using Landsat 5 TM images. The low-pass filtering, median filtering, mean filtering, sum filtering, enhanced adaptive filtering five kinds of spatial filtering methods to do the test. By designing a series of evaluation indexes, the smoothing ability, boundary preservation and enhancement ability of each noise cancellation method are analyzed. Based on the evaluation index and comparison of images, summation filtering (D = 1) and enhanced adaptive filtering are the most effective methods to eliminate Landsat 5 TM image noise in forest land use and forest type classification. Figure 1 Table 2 Reference 29.