Abstract

The C₃ component has much noise and the boundaries of shadow areas are fuzzy in the C₃ component shadow detection algorithm. It is unstable for the algorithm when detecting ground scene information with low saturation. For the deficiency of the detection, the C₃component is normalized and the contrast ratio of the image is enhanced to make the boundaries clearer. For characteristics of the remote sensing image, the Laplacian operator is improved to detect edges precisely with less noise. A shadow detection algorithm which combines the C₃ component and the modified Laplacian operator is proposed. Experimental results show that this algorithm which is simple and effective can not only increase the precision rate of shadow detection, but also reduce the error rate of detecting non-shadows as shadows.

Paper Details