Plug-and-play diffusion priors (PnPDP) have emerged as a promising research direction for solving inverse problems. However, current studies primarily focus on natural image restoration, leaving the performance of these algorithms in scientific inverse problems largely unexplored. To address this gap, we introduce \textsc{InverseBench}, a framework that evaluates diffusion models across five distinct scientific inverse problems. These problems present unique structural challenges that differ from existing benchmarks, arising from critical scientific applications such as optical tomography, medical imaging, black hole imaging, seismology, and fluid dynamics. With \textsc{InverseBench}, we benchmark 14 inverse problem algorithms that use plug-and-play diffusion priors against strong, domain-specific baselines, offering valuable new insights into the strengths and weaknesses of existing algorithms. To facilitate further research and development, we open-source the codebase, along with datasets and pre-trained models, at [https://devzhk.github.io/InverseBench/](https://devzhk.github.io/InverseBench/).

The vehicle-road cooperative perception needs high accuracy and real-time automatic background filtering to separate background objects from foreground objects in complex traffic scenes. Reducing the influence of foreground objects to improve accuracy, and introducing a new framework to improve real-time performance are two main challenges in automatic background filtering. This paper proposes an Automatic Background Filtering method with innovative Frame Selection and Background Matrix Extraction modules (ABF-FSBME) to address these challenges. Firstly, a new space division method with equal hitting probability is proposed to divide the 3D point cloud formed by roadside Light Detection and Ranging (LiDAR), which can reduce the influence of slight LiDAR vibrations. Secondly, the terminal-edge-cloud framework is introduced to balance delay-constrained tasks and computation-intensive tasks in automatic background filtering. Thirdly, a variance-based frame selection strategy with a sliding window mechanism is proposed to select candidate frames with fewer foreground objects. This strategy can reduce the influence of foreground objects in a coarse-grained way. Meanwhile, a new background matrix extraction method is proposed to construct the background matrix. This method can further reduce the influence of foreground objects in a fine-grained way. Finally, based on the extracted background matrix from a cloud server, the edge server can filter the raw frame in real-time. The experimental results show that the proposed ABF-FSBME method has better accuracy than other methods in error rate and integrity rate. Besides, the proposed ABF-FSBME can complete fame filtering within 10ms, and has almost no network delay, so it can satisfy the real-time requirement.

Currently, image copy–move forgery issues have brought increasingly serious problems to the social fields. Many research have been devoted to address the image copy–move forgery issues. However, image copy–move forgery detection (CMFD) is still a challenging problem to image forensics. This paper proposes a coarse-to-fine detection method fusing the superiorities of both keypoint-based and block-based methods. The fusion method gets good geometric invariances of keypoint-based methods and good matching robustness with the invariant moment of block-based methods. In the coarse detection stage, a robust SIFT descriptor is used to extract the candidate keypoints, and then the 2NN test is applied to match the suspicious keypoint couples. The proposed three-pass filtering relying on the Euclidean distance, scaling coefficient ratio, and correlation coefficient of block-based features, remove the false-positive outlier couples. Finally, the scaling coefficient ratio statistics of the remaining keypoint couples get the scaling coefficient ratio between the size of copied/pasted or pasted/copied snippets. In the fine detection stage, the block-based thought relying on the scaling coefficient ratio uses the DAFMT to extract the block-based features of multiple scaling levels. Subsequently, LSH is presented to classify block features and finally indicate the fine forgery snippets. Finally, the morphological operations are presented to indicate the forgery accurately. The benchmark IMD and CoMoFoD image copy–move datasets are used to measure the performances between the proposed fusion method and the state-of-the-art CMFD methods. The numerous experimental results demonstrate that the proposed fusion method achieves nearly the best performances to resist various attacks, especially large-scaling attacks, among the compared state-of-the-art methods.

A new model of hybrid cloud computing architecture based on cloud bus is proposed. The system is based on local private cloud, combined with one or more type(s) of public cloud(s). The internal structures of private cloud and public cloud are the same, including infrastructure and virtualization layer, cloud platforms layer, cloud bus layer, cloud application layer, the management center and storage centers. The layer of infrastructure and virtualization is designed to incorporate the underlying hardware resources into a virtual cluster, providing a variety of virtual resources to the upper layer. The layer of cloud platform is used to run Web applications or services, and carry application-specific development and application integration through its open interfaces. The cloud bus layer, consisting of a control bus, a number of node buses and adapters, is designed to manage and monitor the various services of the cloud platform layer. The proposed model of the architecture can accelerate the migration of the existing IT environments to cloud computing environments, reduce the investment, and make full use of IT resources.