Font Size: a A A

Image Processing And Object Recognition Based On The Cognitive Mechanism Of Visual Information

Posted on:2016-06-04Degree:MasterType:Thesis
Country:ChinaCandidate:X DuanFull Text:PDF
GTID:2348330542973996Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
As we all know,the visual information contributes to the majority human conceiv e information from the outside world.Studying human visual cognitive function to bui ld the visual progress model is a great significance to promote the computer vision pr ogress.The computer vision is a research field which includes plenty of contents and has enormous challenges.There is still a big problem in manipulating the machine to conceive and comprehend the outside though the vision to control themselves get used to the circumstance.In the long run,the ultimate aim of the computer vision is to b uild a system which can compare favorably with the human vision system.It is a big challenge to the researchers to build the visual model inspired by creature visual syst em.In the dissertation,on the basis of this background,we summarize and analysis t he visual model inspired by human vision system,then apply the models in progressin g images and recognizing the objects.The main contents is as fellows:In the dissertation,the physiological property and mechanism of the visual system are analyzed,including the structure of human eyes,the transportation of visual infor mation in visual pass way,the visual information processing areas in brain cortex.Hu man eyes receive the visual information from the outside though the retina.Then the visual information is processed by the visual system.The visual information is transfo rmed into the visual passage in the brain.At last the information is perceived in the b rain.At the end of the last century,the researchers found the synchronous oscillation in the cat's cortex.The visual cortex synchronous oscillation model was established.Then the model was modified.The Pulse Coupled Neural Network was formed.Pulse coupled neural network(PCNN)based on synchronous oscillation in visual cortex neuron has well property in image processing,such as scale invariance strength and distortion invariance.Relying on the nature property of the images,the PCNN has a high real-time performance.Otherwise,The combination between PCNN and the median filtering algorithm achieve a good result in image denoising.The image segment using PCNN is analysed in different parameters.The PCNN is compared with other algotithms in image edge detection.The visual attention is an important mechanism in obtaining the visual informati on.There is a distribution processing when human obtain the visual information.The property,focusing attention on outstanding object when human obtain the visual information,is named visual attention.The domestic and foreign researchers build numerou s models based on visual attention.Several visual attention models are analysed in thi s dissertation and comparing experiments are done on image data base.The visual atte ntion IG models performs excellent in the experiments though analysing the saliency maps.Combining the IG model and image segment algorithm to extract the salient ob jectis pepared for the object recognization in the next step.After using PCNN model to preprocess images,the attention model is applied to extract and separate the target object.Taking advantage of the prevalent local image f eature recognization algorithm to match and recognise the target object.Several algorit hms,such as SIFT SURF ORB,are compared to proved their preporties in local imag e matching.The experiments prove that the ORB has the best performance.Combinin g the visual attention model and the ORB algorithm to constitute a system to recognis e the target object.
Keywords/Search Tags:visual information, PCNN, visual attention, local image feature recognization, image matching
PDF Full Text Request
Related items