Image Analysis System for Optimal Geometric Distribution of RFID Tags Based on Flood Fill and DLT.

A novel image analysis method for improving the reading performance of the radio frequency identification (RFID) system is proposed by arranging the geometric distribution of multiple tags reasonably based on flood fill and direct linear transformation (DLT) in this paper. An image collecting system is projected to collect the image of RFID tags, and an online detection system is designed to measure the reading distance. The proposed system has good performance of instantaneity and robustness, in which tags could be identified rapidly with the tray turning. The 2-D distribution images of tags are processed using the flood-fill technique. Furthermore, the 3-D distribution images of tags are analyzed based on DLT. Subsequently, a tag network is established, in which a tag is described as a node. Finally, support vector machine is introduced to predict the reading distance and corresponding distribution for a given number of tags after training the measured reading distance of RFID tags under various geometric distributions. The results show that the optimal geometric distribution of tags could be obtained by analyzing the images of tags and the predicted reading distance of tags. This method can improve the dynamic performance of multiple tags in the RFID system effectively. [ABSTRACT FROM PUBLISHER]