Abstract: Object-oriented change detection technology has been widely used in the field of high-resolution remote sensing images. As the remote sensing images are affected by imaging conditions such as illumination, atmospheric environment and other factors, the quality of image features also varies. Selecting high-quality features becomes the key of the change detection of remote sensing image at the object level. For the above problems, a change detection method of object-level remote sensing images based on Relief-PCA feature selection has been proposed. In the proposed method, first of all, the original image is multi-scaled to obtain the target object. Afterwards, the spectral features and texture features of the object are extracted. Then a logarithmic ratio method is used to obtain the change vector, and the object features of the original image are filtered and dimensioned through the Relief-PCA feature selection method. Finally, the change vector analysis (CVA) variation intensity map is calculated and generated. The Otsu method is used to conduct the threshold segmentation of the variation intensity map to obtain the final change detection result. Experimental results show that compared with other state-of-the-art methods, the proposed method has higher detection accuracy, lower misdetection rate and lower missed detection rate.

Key words:  remote sensing image, Relief-PCA, change detection, image feature