Bullet-hole image classification with support vector machines
Wenfang Xie, Dongbin Hou, Q. Song · 2000
This paper focuses on the application of support vector machines (SVM) for classification of bullet hole images in an auto-scoring system. In order to automatically calculate the score of a shooter, the bullet-hole images can be classified as one, two or more bullet-hole images. For the auto-scoring system, two main issues are considered. One is to extract important features of bullet-hole images from the target paper; the other is to classify these images into correct classes. A set of essential features of bullet-hole images are discussed and used for the subsequent classification. SVM has been applied to the multi-class classification problem. Experimental results show that both the extracted features and SV learning algorithms are effective and efficient for the project.