The Road Recognition Technology Research Of Mobile Robot

According to the advantages of rich information, wide range of signal detection, and intactinformation acquirement of the computer vision, the road recognition of mobile robot based onvision becomes a key point in the research of mobile robot. As with human visual, robot visionsystem can offer it the image information what is from the environment where it is in. Machinevision is the main information what the robot acquires from the environment around it. It is withthe important meaning in the research of vision system of mobile robot and image processingtheory and method.In this paper, on the basis of the existing mobile car added visual system, including visualsensor (CCD camera), with the function of image acquisition data processor (DSP) and so on,can realize the small car in the process of movement on the image acquisition.Through the comparative analysis, respectively from the two aspects of color and texture ofimage feature extraction. Select image histogram of color feature extraction method, andoptimized through histogram equalization image color information, and respectively in thehistogram of color and three colors of the histogram two ways of image color feature extraction;Image texture feature extraction, and the use of image gray level co-occurrence matrix method,characterization of gray level co-occurrence matrix can clear image texture features. Therecognition of image feature extraction for subsequent image made a very good bedding.This paper uses the support vector machine (SVM), which is the core of the imagerecognition and classification algorithms. Establishing characteristic vector, by a large number ofeigenvector learning supported by every type of image vector, and then realizes the imageclassification.Finally, through software programming, the establishment of image preprocessing, imagefeature analysis and classification of image recognition method of actual framework, discussedin this paper, four kinds of typical road image recognition classification experiment was carriedout.From the experimental results analysis, summed up the application of support vector theoryfour road recognition can achieve high accuracy, which shows that the method is feasible andreliable.