Image Processing and Deep Normalized CNN for the Location Measurement and Reading Distance Prediction of RFID Multitags

In the field of radio frequency identification (RFID) application, the locations of RFID multitags have a great influence on the reading distance of RFID system. In order to improve the reading distance of RFID multitags, a novel image processing and deep normalized convolutional neural network (DNCNN) for location measurement and reading distance prediction of RFID multitags are proposed. First, in view of the complex electromagnetic interference environment, image technology is involved in the experimental platform design. The designed experimental testing platform is utilized to measure the reading distance and collects the images of RFID multitags. Second, due to the existence of motion blur in the acquired images, the hybrid deep priors (HYDP) method is proposed to restore the blurred images. Then, the template matching method is utilized to measure the 3-D locations of RFID multitags quickly and accurately. Finally, DNCNN is proposed to deeply mine the relationship between the RFID tags’ location distributions and corresponding reading distance. The experiments are conducted to verify the effectiveness and superiority of constructed DNCNN. The experimental results show that the relationship between the tags’ location distributions and corresponding reading distance can be well built by DNCNN. The constructed DNCNN can effectively predict the reading distance of RFID multitags. The mean absolute error (MAE), root mean square error (RMSE), and mean absolute percent error (MAPE) of DNCNN prediction results are 0.0377 m, 0.0433 m, and 2.45%, respectively. The proposed DNCNN in this article can find the optimal location distribution of RFID tags corresponding the max predicted reading distance, which can effectively improve the reading performance of RFID system.