结合SFIM和区域填充的对象场景融合

Object-Scene Fusion by Combining SFIM with Region Filling

针对Poisson融合法和现有基于IHS变换的方法存在的问题,在保留IHS变换优势基础上提出了一种结合基于平滑滤波的亮度调制(smoothing filter-based intensity modulation,SFIM)技术和区域填充的综合改进方案.该方案引入SFIM来降低对象与场景融合时的颜色畸变,用色度、饱和度填充解决细节还原系数无法有效增大(对象获得足够融合光照)以及融合对象光照不一致的问题.主观质量评价和客观质量评价均表明该方案不仅可有效解决上述问题,而且融合速度快,算法更加健壮.

Considering the drawbacks of the Poisson fusion technique and an existing object-scene fusion method based on IHS transform, this paper proposes an improved scheme by combining smoothing filter- based intensity modulation （SFIM） with region filling making use of the advantages of IHS transform. The scheme reduces color distortion in fusion of object and scene by introducing the SFIM technique. With the hue and saturation filling, it can also fix the problems that the detail recovery cannot be increased effectively to let object get enough fusion illumination so that the object may appear inconsistent in illumination. Both subjective and objective quality assessments show that the proposed scheme can solve the above-mentioned problems. It is fast and more robust.