一种基于单目移动RGB相机的光照估计方法

A method of light estimation based on monocular moving RGB camera

针对增强现实应用中因光照估计难度大和准确率低造成的虚实光照不一致问题,提出1种基于单目移动相机的全局光照估计算法,通过自适应相机参数选择,将图像各像素作为等效点光源估算光照强度,并结合设备内置传感器进一步计算各点光源位置;通过连通区域分析方法,拟合出对光照影响较大的各类型光源;借助多光源光照计算方法,计算估测光源对虚拟物体的影响.

Aiming at the problem of inconsistent lighting and real lighting caused by the difficulty and low accuracy of lighting estimation in applications such as augmented reality,a global illumination estimation algorithm based on a monocular mobile camera was proposed.By adaptive camera parameter selection,each pixel of the image was regarded as equivalent point light source to estimate the intensity of the light,and the built-in gyroscope and other sensors were used to further calculate the position of each point light source.Through the connected component analysis method,the adjacent point light sources were merged to fit each of the light effects that have a greater impact on the light.With the help of multi-light source lighting calculation method,the impact of light source on virtual objects were calculated and estimated.The experimental results show that this algorithm can effectively estimate the lighting information in the real environment and solve the consistency problem of virtual and real lighting to a certain extent.