Comparing Different Feature Extraction Methods of Pump Dynamograph Based on Support Vector Machine

The paper use two different methods of image moment invariants and wavelet packet-energy entropy (WP-EE) to extract the pump dynamograph’s eigenvector, then making the collected date is training samples and test samples for the support vector machine (SVM) classifier separately, finally comparing the accuracy of two methods. The experimental results show that when choosing classifier based on support vector machine, the method of extracting image moment invariants has more higher accuracy than the other method.