Research Of Saliency Target Detection Method Based On Background Suppression

As a preprocessing module of image analysis and understanding,saliency detection has become the focus of research in this field.Aiming at the problem that the accuracy of the traditional saliency detection method is low under complex background interference and the target boundary is blurred and the significant region is too large.This paper presents a method of detecting the significance of background suppression.Firstly,in the feature extraction stage,the feature is extracted by means of hyper pixel segmentation,and the undirected graph is constructed;Then at the stage of computing the significant,significant value with a priori target joint learning CRFs inference node,and through the hypercomplex Fu Liye transform to eliminate repetition background in the frequency domain,complete the background significant inhibition of target positioning;In the final saliency map generation phase,through the semi supervised pattern ranking introducing airspace on the edge of the background prior to,again to suppress background differences,then significant goals are the final saliency map.The experimental results show that the method proposed in this paper,pedestrians,cars and bicycles target saliency detection in complex background suppression and can accurately lock the specified target position,and get the saliency map of the target complete and clear boundary.Compared with other flow pattern methods,the proposed method improves the evaluation index significantly,and has better accuracy in detecting complex background.