Simple shadow removal using shadow depth map and illumination-invariant feature

Shadows included in images provide useful information for visual scene analysis, but are also factors that negatively affect digital image analysis. Therefore, shadow detection and removal must be considered essential in the preprocessing of the digital image analysis process. In this paper, the shadow region included in the image is detected using an illumination-invariant image whose characteristics do not change even under the influence of various illuminances, and a shadow removal method using the multi-channel gamma correction and a shadow depth map is proposed. In particular, cast shadows include umbra, which is a shadow that is completely obscured by an object that is covered by a light source according to the intensity of light, and penumbra, which is caused by the diffraction effect. In performing gamma correction of these two regions, the shadow was removed by increasing the brightness of the umbra compared to the penumbra region using the shadow depth map generated based on the statistical characteristics of the detected shadow region. As a result of the experiment, it was shown that the shadow removal of the proposed method effectively removes the umbra region in the natural image containing the shadow.