Simple and efficient approach for shadow removal from a single-image
Siham Aqel, My Abdelouahed Sabri, Abdellah Aarab · International journal of imaging and robotics · 2016
In this paper, we present a simple and efficient shadow removal for single natural images. Shadows in image can help and reveal information about illumination sources as well as object’s shape and orientation. However, in computer vision, the presence of shadow will adversely affect the results of some algorithms such as image segmentation, object tracking … Thus, in natural images processing, it is imperative to detect the presence of the shadow and if necessary eliminate it. This work proposes an efficient approach for shadow detection and subsequently elimination in natural images where textural features are used in the detection process while photometric features are used for shadow elimination. We first analyze texture information’s and proximity in terms of similarity between a pair of segmented regions to identify shaded regions. And for removal, we use a Gaussian model to estimate the value of the free-shadow region based on the histogram information. To evaluate the performance of the proposed approach, experimentations are made on several type of shadow.