Method for Detecting Visible Impurity in Oral Liquid Based on Improved SURF Algorithm

Abstract

In order to solve the problems that the visible foreign objects in the large bottle oral liquid are small, the detection speed is low and the single feature would be easy to cause the mis-tracking. In this paper, the speed-up robust features (SURF) algorithm is improved and applied to the detection of visible foreign objects. Firstly, features from accelerated segment test (FAST) detection algorithm is used instead of the Hessian matrix for feature point detection to avoid the extraction of numerous and useless feature points in the edge region. Secondly, two-way fast library for approximate nearest neighbours (Flann) algorithm is adopted for the feature matching to accelerate the matching rate and improve the accuracy of matching. The related experiment shows that the proposed algorithm can accurately match the target and effectively improve the detection speed, which meets the requirements of online detection.