Target Recognition Of Visual Attention Model Based On The Incremental Memory

Visual attention mechanism is a key way to get attention from the natural scene in the region of interest,and the key steps of the limited brain resource to deal with many complex information.With the continuous development and improvement of computer vision,the visual attention in image processing has become a hot research area,especially in the area of image complex area of target identification is becoming more and more obvious.By simulating the human visual attention mechanism,when a natural image can be processed and the salient region of the image can be processed and identified,and the identification process is very time-consuming,and the accuracy rate is very high..At present,the research of object recognition in the visual attention mechanism of human visual system is mainly focused on the bottom-up visual attention model and the top-down visual attention model.From bottom up to the research of visual attention model,some progress has been made,but the research on target recognition is not much,but due to the complexity of human attention,there are some branches of the research.In this paper,we combine the two parts to complete the recognition of the target from the bottom-up visual attention model and the top-down visual attention model.First of all,a linear filter of an input image is acquired by Gauss Pyramid and the Center-surround Difference is obtained.After the independent component analysis,the image is obtained.On this basis,the model is used to carry out the accurate recognition by the top-down visual attention model.At the same time,this model introduces an incremental memory mechanism,which is a kind of mechanism of training learning knowledge about the target object.By training the target object and extracting the target knowledge,the knowledge is introduced into the target recognition model,which makes the model more quickly identify the target in complex scenes.The model uses support vector machine to realize the training work of the model.Through the experimental simulation,the model of the model is introduced into the complex target recognition by combining the bottom-up visual attention model and the top-down visual attention model,improve the efficiency of visual attention,save the time of target recognition,and have a good performance in the identification of complex targets.Therefore,the new model can be applied to the identification of complex targets in natural scenes.