Computerised auto-scoring system based upon feature extraction and neural network technologies

This paper presents an application of pattern recognition for the shooting target paper image. This is a two-stage approach. First, essential features of local images are properly defined and extracted from the target paper i mage. Then the Fuzzy C-Means clustering is carried out for determining the initial parameters and structure of the RBF classifier, which is further en hanced by the hybrid leaning algorithm. Experimental results show that the system achieves satisfactory performance both in terms of error rates of cl assification and learning efficiency.