Hybrid Shadow Restitution Technique for Shadow-Free Scene Reconstruction

Shadows are treated as a noise in computer vision scenario, even though it may found useful in many applications. This research focuses the insignificant shadow restitution methodology to improve the scene visibility and to support the dynamic range reduction. The Hybrid technique combines the physical, geometric, textural, spatial and photometric features for shadow detection. Using feature importance statistics the appropriate criteria is chosen and applied. The experiments over wide dataset prove that the proposed hybrid technique outperforms peer research proposals with the expense of computational cost and time. The output results in a shadow-free, visually plausible high quality image.