全景图像是虚拟现实媒体内容最主要的形式之一，可以为观看者提供360度自由观看的效果。全景图像在采集、拼接、压缩、传输和播放等环节都可能对图像质量造成损害，严重影响观看者的体验，准确地评价全景图像的质量对推进其应用有重要的意义。本文从观看者注意力分布角度，提出一种基于显著性检测的无参考全景图像质量评价方法，可减少弱纹理区域对质量计算的干扰；此外，从全局质量感知并结合注意力分布特性，提出一种基于多特征融合的无参考全景图像质量评价方法，能更好地反映全景图像的颜色及清晰度质量。

Panoramic image, as one of the most important forms of virtual reality media content, can provide a 360 degree free view. Acquisition, splicing, compression, transmission and playback may compromise the quality of the image, greatly influencing the viewer's experience. An accurate assessment of panoramic image quality is of great significance to promote its application. From the perspective of viewer attention distribution, this paper proposes a no-reference panoramic image quality assessment method based on saliency detection, which can reduce the interference of weak texture areas to quality evaluation. In addition, from the global quality perception and combined with the characteristics of attention distribution, a no-reference panoramic image quality assessment method based on multi-feature fusion is proposed, which can better reflect the color and definition quality of panoramic image.