An Algorithm of Object Detection Based on Regression Learning for Remote Sensing Images

Abstract

Saliency detection method is widely used in the field of object detection. It can greatly improve the efficiency of detection by extracting the region that may contain the objects and reduce the processing of background area. It is very difficult to detect the airports in remote sensing image for its changeable structure and complex background. Considering this situation, this paper proposes a object detection method of saliency model based on regression learning, this method extracts salient regions that may contain airports, and then detects salient regions through feature bag model. Our method uses super-pixel segmentation, minimization of absolute shrinkage and selection operator and hierarchical learning theory based on back-propagation. The salient regions that contain potential objects are extracted and sent to the feature bag model for classification. and it can location the final objects. This paper constructs the airport training dataset and test dataset, and the experimental results fully prove the effectiveness and superiority of this method.