Image Saliency Detection Based On Visual Attention Mechanism

The definition of image saliency detection is extracting most attracted attention from images by using image processing methods. Visual attention mechanism refers to the human ability to select and process the area in visual scenes. If salient areas can be extracted from images and videos accurately and computer resources can be allocated to these salient areas, the efficiency of image analysis methods will be improved greatly. This paper studies feature extraction and saliency calculation method according to the basic principle of human visual attention mechanism. The main research works are listed as follows:(1) In view of the problem of unclear regional outline, weak anti-noise ability and heavy computation in traditional image salient region detection, a multi-scales image saliency detection based on block contract is proposed. This method based on Itti’s model, extracts brightness, color and direction as image features under three scales to express the overall characteristics of images and then computes local contrast of image block as the saliency value of the image to reduce the amount of calculation and sensitivity to background noise. Simulation experiment result demonstrates that this method can extract image salient region with clear boundary accurately and quickly.(2) In order to detect saliency object with complete boundary accurately and suppress background in natural scenes, a image saliency detection based on boundary prior is proposed. The original image is first segmented into superpixels using SLIC segmentation algorithm so that the detection result can keep the shape of the objects in the image to the largest extent. Then according to the theory of boundary prior, four boundary prior saliency maps are calculated separately to separate the background and salient object roughly. Finally, the final saliency map which further highlight salient object is generated by regarding the centroid of the boundary prior saliency map as the center of salient object to overcome the problem that the traditional center prior fails to detect the target which deviated from the center of the image. Simulation experiment result demonstrates that this method can uniformly highlight saliency object and effectively suppress the background in natural scenes.(3) In order to reflect salient regions in video sequence accrrately, a spatiotemporal saliency detection integrated with motion characteristic is proposed. First, use SLIC segmentation algorithm to segment each frame into superpixels and extracts the color histogram as the feature. Then, use optical flow vector regional construct to compute temporal saliency, and in view of the global contrast model ignores the characteristics of spatial distribution, compute spatial saliency by joining the color spatial distribution. Finally, use an adaptive fusion strategy to merge the temporal saliency map and the spatial saliency map into final spatiotemporal saliency map. Simulation experiment result demonstrates that this method is able to extract salient target with clear outline in dynamic scenes.