The presence of shadows in Remote Sensing images leads to misinterpretation of objects in several real-world applications, which includes Very High Resolution (VHR) image data from urban areas. Consequently, new methodologies are required to analyze urban data efficiently, due to the great variety of artifacts and shadows formed by elevated objects in the image. In this paper, a novel automatic shadow removal approach is proposed to recover missing information caused by shadows and other obstruction artifacts. First, an automated shadow detection method is applied by computing morphological operations between objects and their surroundings, which are combined with shadow spectral features extracted from a color space model, avoiding, this way, false detections in the shadow-originated mask. Second, to recovering the missing information from the shadow guidance mask, an inpainting-inspired strategy is proposed, which unifies anisotropic diffusion, transport equation and texture synthesis into a robust and concise framework. The performance of our approach is evaluated by taking a WorldView-2 imagery, where it was found that the method achieves an overall accuracy on shadow detection up to 90%, in addition to a low rate of false detection (~20%). Moreover, the designed algorithm outperforms existing recovering techniques, providing high computational performance over VHR satellite images that could be suitable for object recognition, land-cover mapping, 3D reconstruction and, particularly, for developing countries where land use and land cover are rapidly changing with tall buildings/structures within urban areas.