Study On Wheel Shape Online Identification Method And Implementation

With the continuous development of the auto industry, automobile hub for automaticproduction is put forward higher request, and the wheel hub recognition is one of the keysteps in automatic production. Put forward a recognition method of wheel appearance isthe main purpose of this paper, and then completes a wheel recognition system; the hub ofthe eccentricity can also be measured in this system to meet the needs of the wheel hubautomation production. Deep analysis of various aspects of wheel hub recognition systemis studied based on both research of the automobile hub recognition situation at home andabroad and similar image recognition systems.First of all, according to the characteristics of the noise in hub image, and a kind ofpretreatment method is put forward, mainly including the image grey, filter, thresholdsegmentation and smoothing burr.Secondly, characteristics of wheel hub for recognition and the extraction methods ofthe characteristic value are given. The characteristics of the wheel hub that will beextracted include the radius size of the wheel hub, the number of the wheel hub spokesand the sequence values that describe the shape of the spokes. The edge of wheel hubimage is extracted and the radius of outline of wheel round is calculated with the methodof least square in wheel edge image. In binary image, the number of wheel hub spokes isgot by calculation of times of jump between the black and white neighbor pixels. Enoughconcentric circles are created again with hub for the center circle, and then calculates allcircular arcs with white pixel on each concentric circle, the minimum one is considered asthe sequence value. Wheel recognition is started using the characteristics extracted.Again, the eccentricity of wheel hub that is the ratio of the distance between center ofwheel center heart circle and center of outline circle and radius of outline circle iscalculated. The smaller the eccentricity the quality of hub is higher. The entire candidatecircles in wheel image are extracted by means of least square, and then the real targetcircles are got during sorting and filtering. Distance of center of two circles is calculated,and then the ratio is got.Finally, recognition algorithm and calculation method of eccentricity are realized in the Visual C++ 2008 programming environment, and then a wheel identification system isdesigned and implemented. Better result is showed through experiment, which the methodis reliable.