Invertible networks have various benefits for image denoising since they are lightweight, information-lossless, and memory-saving during back-propagation. However, applying invertible models to remove noise is challenging because the input is noisy, and the reversed output is clean, following two different distributions. We propose an invertible denoising network, InvDN, to address this challenge. InvDN transforms the noisy input into a low-resolution clean image and a latent representation containing noise. To discard noise and restore the clean image, InvDN replaces the noisy latent representation with another one sampled from a prior distribution during reversion. The denoising performance of InvDN is better than all the existing competitive models, achieving a new state-of-the-art result for the SIDD dataset while enjoying less run time. Moreover, the size of InvDN is far smaller, only having 4.2% of the number of parameters compared to the most recently proposed DANet. Further, via manipulating the noisy latent representation, InvDN is also able to generate noise more similar to the original one. Our code is available at: https://github.com/Yang-Liu1082/InvDN.git.
In this study, a real-time remote monitoring and fault diagnosis method has been developed based on the Internet of Things (IoT) frame perception, and successfully applied to a mine hoist system. The proposed method combines the sensor technology, online monitoring technology, wireless transmission technology, and fault diagnosis technology. The basic structure of the traditional IoT comprises a perception layer, a network layer, and an application layer, the proposed structure contains an additional middleware layer between the network layer and the application layer. This four-layer system is used in a mine hoist remote monitoring and fault diagnosis framework to process heterogeneous multi-source information. The sensors and parameters are connected in the perception layer, the characteristic parameters are obtained using the configuration software, and the mine local area network is saved to the data server, thereby synchronizing real-time data in the local area network. The network layer utilizes mature Internet and long-distance wireless transmission communication technologies, whereas the middleware layer comprises of a Service-Oriented Architecture (SOA)-based IoT data processing framework that integrates the multi-source heterogeneous data. Further, the fault diagnosis method is analyzed and verified based on the gray association rules. In the application layer, a human-computer interface is used for the remote monitoring and diagnosis of the mine hoist and to provide the diagnosis results as feedback to the user. The results using the aforementioned analyses are applied to the remote monitoring and diagnosis of a mine hoist system. In this study, experimental tests are conducted in this study to significantly improve the fault monitoring, diagnostic capabilities, and reliability of the mine hoist system, indicating the good application good prospects of the proposed method.
Opportunistic network is a kind of network which does not have a definite path from the source node to the destination node. It can only communicate when nodes meet each other. According to the classic routing protocol in opportunistic network, a routing protocol is proposed which consider the duration of historical connections with acknowledgement mechanism (DoC-ACK) in this paper. The simulation results show that the proposed routing protocol can substantially improve the message delivery rate and significantly reduce the overhead rate without changing the transmission delay basically.
Accurate, fast, and anti-jamming extraction of frame is the primary tasks of the indoor scene reconstruction. The existing frame extraction methods are complex, and the extraction results are in low fitting degree with the real indoor frame. Extracting frame structure from the indoor scene is the primary task of indoor scene understands. The illumination, occlusion, and other interference make the computer automatically identify the framework full of challenges. For the five-sided indoor scene image, although the vanishing points in three directions are not all necessarily detected, there must be a vanishing point located inside the frame intersected by feature lines of other two directions. For the indoor scene frame structure detection method proposed in this paper, the feature lines in the direction of three vanishing points are important clue and foundation. According to the candidate corner vertices detected in the previous step, corner vertices of the frame will be determined.
3D digital library is a hotspot in recent years. With the continuing development of 3D graphics technology, 3D model as the fourth-generation media has been one of the digital resources in digital library. This article analyzes the progress of two directions: 3D model indexing and compression. At last, we propose a simple frame of 3D model indexing and compression in digital library.
A new method for detecting roads in SAR images is presented in this paper. The method consists of four parts: preprocessing, Hough transform, genetic algorithm and region growing. In the preprocessing step, morphological operations are employed to remove many details in binarized SAR images. The Canny operator is applied to detect image edge, and small areas are deleted by using morphological area open operation. Hough transform is employed to extract lines. Genetic algorithm (GA) is used to search and connect road segments. Region growing approaches are utilized to extend road segments. The experimental results show that the proposed method can better detect roads in SAR images.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)