Shadow Detecting Using Mathematical Morphology and Smirnov Test
Chao Xing, LI Yan-jun, Ke Zhang · 2010
An algorithm combining intensity information and geometric features is introduced in order to detect cast shadows in a gray level image. A simple connected candidate shadow region and a corresponding region are selected by setting gray level thresholds, and neighbor-matching regions are constructed with mathematical morphological algorithm. Shadow-non-shadow region pair is obtained from the result of Smirnov test for statistical features of candidate neighbor-matching region pairs, thus shadow regions are detected by selecting one with relatively lower intensity average from the matched two regions. Experimental results show the effectiveness of the algorithm for cast shadow detecting in gray level images.