Research On Key Technologies Of Vehicle Recognition Algorithm Based On Regional Salient Feature

Vehicle recognition technology which belongs to the computer vision, imageprocessing and pattern recognition technology plays a very important role in the intelligenttransportation field. Vehicle license plate recognition system is an important part of thevehicle recognition technology. Vehicle logo is also one of the marks and vehicle logorecognition can improve the recognition rate of the vehicle recognition. This paper dosome research on vehicle license plate location and vehicle logo recognition, and proposesome novel algorithm.According to license plate edge color collocation feature and character structurefeature, a vehicle license plate locating algorithm based on edge-color pair feature andstroke traversing is proposed. First of all, feature data of all the color edge collocation iscollected artificially. Edge-color pair feature coverage model is constructed throughmachine learning, and then the constructed model is used and the plate candidate area isshaped after morphology processing by using prior information. Finally, edge traversinginformation is scanned in the license plate candidate area and the license plate candidatearea is validated using the license plate overall regional edge distribution characteristics.Experimental results show that the proposed method is reliable.By considering vehicle license plate locating as salient region detection, a vehiclelicense plate locating algorithm based on salient feature and character feature is proposed.First of all, the color feature and energy feature is considered as salient feature, and thenthe feature maps are combined. After the binarization of the map, plate candidates areobtained. The character edges are detected in the input image at the same time and theedges that not belong to character are removed. Finally, according to character strokefeature and prior knowledge, the real license plate is confirmed. The better experimentalresults are obtained.A vehicle logo location and recognition algorithm based on edge distribution featuresand phase congruency is proposed. First of all, the front area of the vehicle is segmentedby making use of the vehicle logo and license plate relative position and texture information. The logo is roughly located in the front area of the vehicle, and then thevehicle logo can be exactly detected by making use of the edge distribution features toinhibit the influence of background. In the recognition module, considering the influenceof background, phase congruency is utilized. Finally, the logo can be confirmed by HOGpyramid feature. The test results show that the algorithm is feasible.