Automatic Geometric Correction of Complex Sea Condition Remote Sensing Image Based on Decision Tree Classification

Abstract

Abstract The geometric correction of ocean remote sensing image is a prerequisite for its data application. In this paper, to solve the problem that the sea island is sparse, cloud interference is severe, the control point is difficult to obtain, an automatic correction technique based on decision tree classification is proposed. In this paper, the image is processed by the method of super-pixel segmentation first. Then, the spectral and texture features in the superpixels are selected, including the energy value, the entropy and the correlation value of the gray level co-occurrence matrix and the normalized water index. Finally, the tree image classification model is used to classify the image superpixels, and the clear sky area which will be matched directly with the reference image can be extracted. Through the template matching and polynomial geometric model, the geometric correction of the remote sensing image is automatically corrected. Through the experiment of Landset8 OLI_TIRS image, compared with the classification results of the other two classification methods, the final precision is better than the other two methods. Therefore, the technical process proposed in this paper can be applied to the geometric correction of complex sea condition remote sensing images.