License plate detection using channel scale space and color based detection method

Vehicle detection is to detect vehicles that are being developed to provide traffic counting, vehicle classification. In existing method, vehicles are identified by using channel scale space of Red, Green, Blue (RGB) images, taking license plate as a feature. The union of edge image is taken by combining red region, green region, blue region and grey region. The shape of License Plate (LP) is referred as candidate shape. Based on the edge points there are few issues in the candidate shapes. They are outliers, open shape and fragmentation. The outliers are removed by the removal of outer outlier method which is to remove the unwanted projections in the candidate shapes. In proposed method, color based detection is used along with channel scale space technique. In this color based detection method, CIE-XYZ color model is used. By calculating the aspect ratio, the license plate is detected to identify the vehicles. The accuracy of this method is 94.23%.