多失真混杂的无人机图像质量盲评价

发布时间：2018-06-07 15:46

本文选题：无人机图像 + 多失真混杂　；参考：《中国图象图形学报》2017年01期

【摘要】：目的面向多失真混杂的图像质量盲评价问题目前仍然是计算机视觉领域具有挑战性的工作之一,无人机图像受成像条件影响混杂多类失真,图像质量的准确评价是其效能发挥的关键环节。为此,引入并改进了基于自然场景统计的距离度量评价模型,提出多失真混杂的无人机图像质量盲评价方法。方法从图像的结构性、信息完整性和颜色性3个不同的角度研究并提取了与无人机图像质量敏感的特征因子集;以实拍标准测绘图像库为原始图像获得MVG特性参数作为度量基准解决了盲评价中缺乏训练集的问题;构建了以实飞图像为样本的无人机图像质量数据库(UAV image set),为相关问题的研究提供数据集和评价参考。结果针对所构建的数据库,本文算法在主客观一致性、算法运行时间上与其他算法进行了对比实验。相比较其他经典算法,本文算法的主客观一致性较高,达到了0.8以上,运行时间较快,过到1.2 s。此外本文还给出了块大小对算法影响以及单特征对图像的评价结果,证明算法选择的图像块大小和图像特征符合质量评价的需要。结论针对无人机图像所包含的多失真构建质量评价综合模型,该模型可满足无人机图像质量需求。
[Abstract]:Objective Blind evaluation of image quality for multi-distortion hybrid images is still one of the challenging tasks in computer vision field. UAV images are affected by imaging conditions. The accurate evaluation of image quality is the key to its effectiveness. This paper introduces and improves the distance measurement evaluation model based on natural scene statistics, and proposes a blind evaluation method of UAV image quality with multi-distortion hybrid. Methods the feature factor sets which are sensitive to UAV image quality were studied and extracted from three different angles of image structure, information integrity and color. The problem of lack of training set in blind evaluation is solved by using the MVG characteristic parameter obtained from the original image as the measurement benchmark. A UAV image setter database based on real flight images is constructed, which provides data sets and evaluation references for the research of related problems. Results based on the database, the algorithm is compared with other algorithms in subjective and objective consistency and running time. Compared with other classical algorithms, the subjective and objective consistency of this algorithm is higher than that of other classical algorithms, reaching more than 0.8, and the running time is faster than 1.2 s. In addition, the effect of block size on the algorithm and the evaluation result of image by single feature are given. It is proved that the block size and image feature selected by the algorithm meet the needs of quality evaluation. Conclusion the synthetic model of UAV image quality evaluation is constructed for the multi-distortion of UAV image, and the model can meet the requirements of UAV image quality.
【作者单位】：解放军陆军军官学院;
【基金】：安徽省自然科学基金项目(1208085MF97)~~
【分类号】：V279;TP391.41

【相似文献】