敦煌壁画数字图像拼接评价模型研究

摘要

针对敦煌壁画数字图像拼接后存在的评价问题,系统地分析了图像拼接质量客观定量评价的各种参数和方法,结合敦煌壁画数字图像的特点,提出了一种较完善的图像拼接结果评价标准模型.进而分别用Harris算法、SURF算法对数字图像进行拼接,并对拼接后的图像采用评价标准模型进行评价.通过大量实验证明,这一模型能更准确地反映人眼的主观感受.由于针对不同的场景,图像的各个指标都不尽相同,所以很难指定一个统一的指标标准,针对这一难题,实验中还添加了“自定义设置”,可以时各个指标进行调节,以适应不同的场合.%Aiming at the evaluation problems existing in digital image mosaic for Dunhuang frescoes, a evaluation model of more sophisticated digital image mosaic results is presented in combination with the characteristics of digital images of Dunhuang frescoes based on analyzing various parameters and methods for evaluating the image mosaic quality objectively and quantitatively. The digital images were stitched by using Harris algorithm and SURF algorithm respectively. The stitched images were evaluated with the evaluation model. A lot of experiment results show that this model can reflect the eye's feeling more accurately. As for different scenes, viarious indexes of an image are different, so it is difficult to specify a uniform standard. A "Custom Setting " that can regulate every index to adapt to different scenes was added during the experiment to solve this problem.