Comparative Experimental Study of UAV Image Enhancement for Digital Geomorphology Simulation of Zhongshan Crater
With the rapid advancement of low-altitude UAV technology,small consumer UAVs carrying optical sensors show the advantage of fast and flexible acquisition of high-resolution image data.However,due to the limitations of shooting equipment,environmental conditions and natural illumination,the image acquired by UAV often has problems such as noise,blurring and lack of contrast,which restricts the accuracy and reliability of digital geomorphology simulation.To further explore ways to improve the processing of UAV image data quality,this study studied the UAV image enhancement comparative experiment on the digital landform simulation of Zhongshan crater,carried out a comprehensive image quality evaluation index determination for the UAV image preprocessing technology,and conducted a comparative analysis of image enhancement processing using a variety of algorithms.The Retinex algorithm enhancement technology suitable for UAV image preprocessing in this study is optimized by experiments.Direct gray transform,Mask method,Wallis filter and four image enhancement pro-cessing methods based on Retinex were studied and compared.Statistical image parameters such as information entropy,variance and average gradient were used for quantitative image quality evaluation,and the qualitative results of visual interpretation were checked and verified.Therefore,the SIFT algorithm is further used to com-prehensively compare the extraction accuracy of the edge features of the enhanced image.The empirical compara-tive study shows that the Retinex algorithm can significantly enhance shadow removal and brightness correction,and the objective evaluation index of image feature extraction results enhanced by the Retinex algorithm can im-prove the ratio.Among them,the OA value increased by 4.1012％and the Kappa value increased by 0.0635.
NETL
NSTL
万方数据
