Research On Salient Region Detection Based On Structured Learning

Salient region detection can simulate the human visual system to detect the salient region of the image rapidly,and it can improve the efficiency of image processing.Meanwhile,salient region detection has been widely used in the field of computer vision,such as target recognition,image compression and coding,image retrieval,scenario analysis,etc.This paper firstly analyzes the research background and the research status of salient region detection,then describes the basic concept of visual attention mechanism,the two kinds of visual attention model and the theoretical basis of visual attention model.Secondly,this paper introduces the random forest method from three aspects which include the decision tree and node splitting rules as well as the construction of the random forests,and lists some advantages of random forests.Then this paper mainly introduces three salient detection methods based on supervised learning,includes: salient object detection via bootstrap learning,contextual hypergraph modeling for salient object detection,and salient region detection via high-dimensional color transform.Thirdly,the paper proposing a salient region detection method based on structured learning,appling a structured learning method to salient region detection.Firstly,we get a fixed rectangular region randomly from the local image which includes the labellings,and record the corresponding structured labels.Then,we build a collection of decision trees by using the regional features which includes the structured labels.Finally,we capture the final saliency map by using the supervised learning approach.Experiments show that our method can detect the salient objects accurately.Finally,the full text is summarized and prospected.