A novel method for extracting saliency of ground target based on the knowledge of target region
Hao Wang, Wei Haiping, Bochuan Zhang · 2014
A novel interest region extraction algorithm is proposed, which is based on the knowledge of target region, combined with biological vision mechanism. Depending on the knowledge of target, the algorithm preprocesses the image by applying lateral inhibition filter. Then build the saliency model of target by using structural pattern. The algorithm avoids high-dimension features calculation of pixel and its local neighborhood, therefore greatly enhances the operational speed. Otherwise, the saliency model built through target region knowledge has higher performance and is more reliable than the traditional top-down saliency calculation.