Research On Bionic Perception Based On Human Visual Mechanism

As an extension of human tools,machines have not only changed the way of human production and manufacturing,but also promoted the progress of human society.At present,the manufacturing environment of the machine is increasingly complex and difficult to process.It is difficult for traditional machines to deal with this dynamic and complicated nonlinear state.The main problem is the lack of information independent judgment,reasoning and decision-making ability.For this reason,domestic and foreign scholars regard machine intelligence as the main research object,and the core foundation is that the machine has certain intelligent perception ability.The perceptual method of multi-sensor information has limitations when it perceives complex spatial information.Although the traditional visual perception method can get rich information,it also leads to information redundancy and information processing burden.Against the above,in recent years on the research achievements of cognitive mechanism,has carried out based on human cognitive mechanism study on the method of combining bionic perception and machine vision,in order to realize intelligent machines intellisense,providing essential intellisense method,to enhance the machine according to the complex manufacturing environment'autonomous perception,cognition and decision-making ability,to intelligent machines,intellisense provides a new thought of theoretical research,and provide intelligent application in the field of machine vision technology and method.To this end,Human vision system and biological working mechanism are analyzed.Combining the cognitive science of human brain to study visual perception processing method and operation mechanism.Especially the human attention mechanism of bottom-up and top-down mechanism,reveals the human eye visual perception from the main process of generation,transmission,processing and general regulation of brain pathways of visual information processing,the rule can be used to build visual perception from the notice,the relationship among perceived memory,is proposed in this paper based on the human eye vision bionics perception of machine vision method provides solutions.For the human eye visual attention to the problem of mechanism of optimal image acquisition,An image quality evaluation method based on Siamese network and convolutional neural network is proposed,and the optimal image parameter optimization model of machine vision system is constructed.Based on visual attention mechanism of image processing problems,puts forward a primary feature extraction based on two-dimensional Garbor filter bottom-up visual attention model and based on eye movement data and top-down visual attention model of support vector machine(SVM),the image information processing model of bionic vision perception is obtained.Adopt a migration based learning goals and tasks of feature extraction and classification recognition,the method of combining two goals and tasks of visual attention model to extract feature,realized the goals and tasks based on bionic perception of human eyes.The design and development of simulation test system for machine vision bionic perception is carried out.The test analysis of the defect detection of strip steel was carried out,and the defect spot detected by microsilicon steel was 6?6 pixels,The actual defect size was90?90um~2.Further verify the feasibility of the bionic perception method and model in the realization of visual attention target search.The innovative idea of machine visual perception based on bionic perception of human eyes has some practical significance for promoting intelligent machine to realize autonomous perception.