基于熵权赋权法的灰关联技术的视频质量评价体系研究

目的为了满足视频服务系统根据不同的需要提供不同质量的视频,并且能够识别和量化质量损失,使得视频质量能够得到控制和提高,以满足不同领域的要求。方法采用基于熵权赋权法的灰关联技术构建了一套视频质量评价体系,分为视频质量评估系统和灰关联评估模型2个部分。其中视频质量评估系统包含主观评价和客观评价,主观评价采用标准ITU-T P.910中定义的DCR主观评价方法,客观评价中包含峰值信噪比、结构相似度、清晰度—点锐度算法、四元数奇异值4个指标算法。结果多种方法结合进行综合评价优势明显,解决了主观意识和单指标衡量的局限性问题,能够全面准确的评价实验视频,且与人眼相关性高,客观的反映了视频序列的真实质量。结论最终的综合评价结果证明,文中的视频质量评价体系构建合理,方法选择正确,思路可行,达到了较好的评价效果,对视频质量评价的研究具有一定的指导借鉴意义。

Video Quality Evaluation System of Gray Correlation Technique Based on Entropy Weighting Method

This study was aimed to enable the video service system to provide different video quality according to different needs, and be able to identify and quantify the loss of quality, so that the video quality can be controlled and improved, in order to meet the requirements of different areas. A set of video quality evaluation system was constructed by using grey correlation technique based on entropy weight method, and divided into video quality assessment system and grey correlation evaluation model. The video quality evaluation system included subjective evaluation and objective evaluation. The subjective evaluation adopted DCR subjective evaluation criteria as defined by the ITU-T P.910 method, and the objective evaluation included peak signal to noise ratio, point sharpness, structural similarity and quaternion singular value as four index algorithms. The comprehensive evaluation combing multiple methods had obvious advantages. It solved the limitations in subjective and single-index evaluation, and could evaluate the experimental videos in a comprehensive and correct manner, with high correlation with human eyes and objective reflection of the real quality of video sequences. The final comprehensive evaluation results proved that the video quality evaluation system in this paper was reasonably constructed, the method were correctly selected, and the method was feasible, achieving good results, which has a certain guiding significance to the study of video quality evaluation .