State-of-the-art radio frequency identification (RFID)-based indoor autonomous vehicles localization methods are mostly based on received signal strength indicator (RSSI) measurements. However, the accuracy of these methods is not high enough for real-world scenarios. To overcome this problem, a novel dual-frequency phase difference of arrival (PDOA) ranging-based indoor autonomous vehicle localization and tracking scheme was developed. Firstly, the method gets the distance between the RFID reader and the tag by dual-frequency PDOA ranging. Then, a maximum likelihood estimation and semi-definite programming (SDP)-based localization algorithm is utilized to calculate the position of the autonomous vehicles, which can mitigate the multipath ranging error and obtain a more accurate positioning result. Finally, vehicle traveling information and the position achieved by RFID localization are fused with a Kalman filter (KF). The proposed method can work in a low-density tag deployment environment. Simulation experiment results showed that the proposed vehicle localization and tracking method achieves centimeter-level mean tracking accuracy.
Many intelligent perception systems are progressively approaching people’s daily lives. Device-free localization (DFL), as a technology that can achieve positioning without the target being equipped with any equipment, has broad development prospects. RFID-based radio tomographic imaging (RTI) method has attracted much attention because of its low cost, easy deployment, and strong real-time performance. However, RTI, as a real-time imaging method that uses the line-of-sight path information and analyzes the received signal strength behavior, cannot effectively resist multipath interference. As the number of targets increases, the positioning performance using RTI method will be hit. In response to this problem, this paper proves the feasibility of applying static reflection multipath to device-free localization, and proposes a new training-free multipath enhancement (TFME-RTI) method using known spatial information. By analyzing the static multipath superposition model and target interference characteristics, a transformation model is established to take advantage of the multipath effect to enhance the positioning performance. Experiments demonstrate that the TFME-RTI method has better positioning performance and robustness in two typical indoor scenarios.
The device-free localization (DFL) can locate the target without requiring the target to carry any radio equipment, which has a broad application prospect in the field of indoor positioning. Radio frequency identification (RFID), as a key technology for IoT implementation, has been widely used in objects identification and tracking. RFID-based radio tomographic imaging (RTI) method has received extensive attention due to its low cost, excellent real-time performance and easy to deployment. However, the indoor multipath effect, which leads to unpredictable variations of the received signal strength (RSS), especially in multi-target positioning. It is still a challenging problem to recognize the locations of multiple targets effectively in the complex indoor environment. In this paper, a novel method of intersection multidimensional feature analysis is proposed to solve this problem. The target impact region is introduced to extract the effective intersections, and then the multidimensional feature information is analyzed to determine the region where the real target may exist. The location estimation of the target is obtained by assigning appropriate weights to each intersection in the region. Extensive experiments are conducted with results are demonstrating that the method achieves high localization accuracy and localization stability in different indoor scenarios, while maintaining low computational complexity.
26 June 2024 in Signal, Image and Video Processing
Due to the non-uniform illumination conditions, images captured by sensors often suffer from uneven brightness, low contrast and noise. In order to improve the quality of the image, in this paper, a multi-path interaction network is proposed to enhance the R, G, B channels, and then the three channels are combined into the color image and further adjusted in detail. In the multi-path interaction network, the feature maps in several encoding–decoding subnetworks are used to exchange information across paths, while a high-resolution path is retained to enrich the feature representation. Meanwhile, in order to avoid the possible unnatural results caused by the separation of the R, G, B channels, the output of the multi-path interaction network is corrected in detail to obtain the final enhancement results. Experimental results show that the proposed method can effectively improve the visual quality of low-light images, and the performance is better than the state-of-the-art methods.
With the rapid development of the online shopping, the pursuit of outfit compatibility has become a basic requirement for an increasing number of customers. The existing work on outfit compatibility prediction largely focuses on modeling pairwise item compatibility without considering modeling the whole outfit directly. To address the problem, in this paper, we propose a novel hypergraph-based compatibility modeling scheme named OCPHN, which is able to better model complex relationships among outfits. In OCPHN, we represent the outfit as a hypergraph, where each hypernode represents a category and each hyperedge represents the interactions between multiple categories (i.e., they appear in the same outfit). To better predict outfit compatibility, the hypergraph is transformed into a simple graph, and the message propagation mechanism in the graph convolution network is used to aggregate the neighbours’ information on the node and update the node representations. Furthermore, with learned node representations, an attention mechanism is introduced to compute the outfit compatibility score. Using a benchmark dataset, the experimental results show that the proposed method is an improvement over the strangest baselines in terms of accuracy by about 3% and 1% in the fill-in-the-blank and compatibility prediction tasks, respectively.