Studies On The Object-Oriented Change Detection Methods Of Remote Sensing

Change detection has always been one of the main directions of remote sensing technology application. With the rapid improvement of remote sensing technology, the spatial resolution of remote sensing image increases constantly, spatial characteristics and details of scenery become more abundant. In the analysis of high resolution remote sensing image, on the one hand, traditional analytical methods based on image pixel only consider the spectral information of the surface features, and exhibit a significant shortcomings. On the other hand, object oriented image analysis technique considers both spectral information and spatial information, showing a great advantages in analysis of high resolution remote sensing image. It’s more closer to human thinking mode when object oriented image analysis technique is applied in remote sensing change detection. In theory, it can greatly improve remote sensing change detection effectiveness and accuracy. Therefore, this direction become one of the cutting-edge issues in recent years. In this paper(1) The methods of change detection and object oriented image analysis are summarized. Depending on whether the spatial information of surface features are used, remote sensing change detection methods are divided into two categories: the detection method that only consider spectral information and method that taking into account both spectral information and spatial information. Pros and cons of each method are compared. General Process for object oriented image analysis is summarized;(2) Optimal segmentation scale problem in remote sensing is discussed in depth. Evaluation index Espindola is improved based on the analysis of image segmentation quality evaluation methods. Optimal segmentation scale quantitative determination is studied experimentally using multispectral remote sensing image. Effectiveness and practicality of the method are proved by experimental results.(3) Aiming at the pseudo-changing area caused by inconsistent segmentation, a "second division" image segmentation strategy is proposed. Change detection experiment is conducted to verify its validity.