A novel method based on Change Vector Analysis to improve results of Change Detection Process

The change detection using satellite images is one of the main fields of remote sensing researches. Numerous errors in the images are one problem with the use of satellite imagery. Errors caused by surface illumination, are the main problems in the process of change detection. Therefore, in this study, in order to reduce errors in the results of the detection of changes using change vector analysis, a simple, yet effective method to reduce errors caused by topography is proposed. After applying Tasseled Cap Transform on images, the proposed method compare the angle between direction of change vector with the direction of vector of the pixel in the base image using spectral space, and then calculates the angular threshold. The area under the ROC curve, the Probability Detection and False alarm of proposed method are 0.970, 0.97 and 0.32 respectively.