四种图像相似度算法对于气溶胶时空演变图特征的相似度效果分析

本文提出了一种基于不同相似度算法的针对气溶胶时空演变图中显著特征的查找对比方法。对气溶胶时空演变图中的显著特征进行识别，通过对同一幅气溶胶时空演变图进行平移操作来模拟下一时间段的气溶胶时空演变图，借助相似度算法寻找下一时间段内图像中与之对应的特征，借用相似特征中心位置在两幅图像中的坐标差与平移像素数进行比较来确定所运用的相似度算法的优劣。设计了仿真分析实验，通过实验测试对已知移动不同像素数的图像进行分层解算，对不同相似度算法的解算效果进行了全面的比较。通过仿真对比的结果，以及在与实际测风塔数据进行比较后发现，四种相似度算法中，Ahash （Average Hash）算法和Dhash（Difference Hash）算法表现优异且准确性波动较小，上述结论可为相似度算法在气溶胶密度图谱中的特征查找对比方法提供参考。

英文摘要:

In this paper, a method based on different similarity algorithms is proposed to find and compare the salient features in the spatiotemporal evolution map of aerosols. The salient features in the aerosol spatio-temporal evolution map are identified, and the aerosol spatiotemporal evolution map of the next time period is simulated by shifting the same aerosol spatio-temporal evolution map. The similarity algorithm is used to find the corresponding features in the image of the next time period, and the coordinate difference of the similar feature center position in the two images is compared with the number of translation pixels to determine the advantages and disadvantages of the similarity algorithm used. The simulation analysis experiment is designed. Through the experimental test, the images with known moving different pixel numbers are hierarchically solved, and the calculation effects of different similarity algorithms are comprehensively compared. Through the results of simulation comparison and comparison with the actual wind tower data, it is found that among the four similarity algorithms, Ahash algorithm and Dhash algorithm perform well and the accuracy fluctuates less. The above conclusions can provide a reference for the feature search and comparison method of similarity algorithm in aerosol density map.

查看全文查看/发表评论下载PDF阅读器