A new method on similarity of fingerprints and its critical value determination

The calculation about similarity of fingerprints and its critical value determination is an important approach to control the quality of traditional Chinese medicines (TCM) in the related study. There are many traditional calculation method to compute similarity, such as cosine of the angle, pearson correlation coefficient, euclidean distance, and so on. However, it is found that these traditional methods are not sensitive enough, so that they can not distinguish the difference among many TCM samples. And the critical value is always given subjective, so it is lack of objectivity. Basing on two previous problems and taking advantage of cosine of the angle and the idea of support vector machine(SVM), we propose a mew method: Linear measure, which is used to measure the similarity between two fingerprints. Then we give a way to the determination of critical value utilizing resampling method and empirical distribution function. At last two examples show linear measure has much more merits than traditional calculation method, and the final example proves the feasibility of the way to the critical value determination.