The pressed protuberant character is a reflectorized character on the difference of reflectance.The characteristic of the character image differs completely with a general character based on the chromatic difference of background and foreground.A new method of quality inspecting for the pressed character based on the radial basis function neural network(RBF-NN) is presented.The method proposes the direct gray-scale feature extraction on the ring and radial projection algorithm.The results show that the method keeps the integrity feature of the protuberant character information dramatically and takes on a higher performance of anti-jamming,and improves the inspecting speed,moreover abandons the complex image binarization algorithm.Its speed is 240 characters per minute,and the accuracy is 98.77%.It satifies the requirements of the quality inspecting on-line system for the metal tag presser.