Recent advancements in deep learning, particularly through Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs), have led to significant progress in nighttime image deraining. However, current architectures still struggle to strike an optimal balance between computational efficiency and restoration performance. Moreover, existing methods often fail to fully exploit the intrinsic characteristics of low-light conditions and inadequately model the interaction between rain and illumination. To overcome these challenges, we propose NDMamba, a dual-prior-guided state-space model that addresses nighttime deraining by incorporating degradation cues related to both lighting and rain distribution. Inspired by the Retinex theory, which suggests that rain streak distribution is influenced by the reflectance component of a scene, we propose a Prior Extraction Module (PEM) to jointly model lighting conditions and rain degradation. Furthermore, we design a Prior-Guided Mamba Block (PGMB), which comprises a Lighting-Adaptive Vision State-Space Module (LVSSM) that incorporates illumination priors, and a Rain Distribution Guidance Module (RDGM) to enhance local features in a more refined manner. Extensive experiments demonstrate that NDMamba outperforms state-of-the-art methods on both synthetic and real-world benchmark datasets. Our code is publicly available at https://github.com/tandaily/NDMamba.

This article describes how geographic information systems (GISs) can enable, enrich and enhance geospatial applications and services. Accurate calculation of the similarity among geospatial entities that belong to different data sources is of great importance for geospatial data linking. At present, most research works use the name or category of the entity to measure the similarity of geographic information. Although the geospatial relationship is significant for geographic similarity measure, it has been ignored by most of the previous works. This article introduces the geospatial relationship and topology, and proposes an approach to compute the geospatial record similarity based on multiple features including the geospatial relationships, category and name tags. In order to improve the flexibility and operability, supervised machine learning such as SVM is used for the task of classifying pairs of mapping records. The authors test their approach using three sources, namely, OpenStreetMap, Google and Wikimapia. The results showed that the proposed approach obtained high correlation with the human judgements.

There is a tremendous number of geospatial data becoming available and there are numerous methods for getting access to geospatial sources. However, diverse geospatial data sources lead to various extracting ways. Users without adequate background knowledge are unable to fully exploit the growing amount of geospatial data. Our work provides users with a uniform interaction paradigm to tackle the variant retrieval problems in a universal way. Unifying the retrieval methods requires the ability to invoke processing algorithms by a uniform interaction paradigm. Our approach, implemented in a tool called Karma, encapsulates these algorithms as Web Services. As different sources present diverse non-semantic geospatial data, users cannot fully understand the extracted geospatial data. We have defined a general ontology to align and semantify the retrieved data. In addition, we present the principles underlying our approach, and several running examples are given to demonstrate the feasibility and effectiveness of our prototype.

Since large amount of geospatial data are produced by various sources and stored in incompatible formats, geospatial data integration is difficult because of the shortage of semantics. Despite standardised data format and data access protocols, such as Web Feature Service (WFS), can enable end-users with access to heterogeneous data stored in different formats from various sources, it is still time-consuming and ineffective due to the lack of semantics. To solve this problem, a prototype to implement the geospatial data integration is proposed by addressing the following four problems, i.e., geospatial data retrieving, modeling, linking and integrating. First, we provide a uniform integration paradigm for users to retrieve geospatial data. Then, we align the retrieved geospatial data in the modeling process to eliminate heterogeneity with the help of Karma. Our main contribution focuses on addressing the third problem. Previous work has been done by defining a set of semantic rules for performing the linking process. However, the geospatial data has some specific geospatial relationships, which is significant for linking but cannot be solved by the Semantic Web techniques directly. We take advantage of such unique features about geospatial data to implement the linking process. In addition, the previous work will meet a complicated problem when the geospatial data sources are in different languages. In contrast, our proposed linking algorithms are endowed with translation function, which can save the translating cost among all the geospatial sources with different languages. Finally, the geospatial data is integrated by eliminating data redundancy and combining the complementary properties from the linked records. We mainly adopt four kinds of geospatial data sources, namely, OpenStreetMap(OSM), Wikmapia, USGS and EPA, to evaluate the performance of the proposed approach. The experimental results illustrate that the proposed linking method can get high performance in generating the matched candidate record pairs in terms of Reduction Ratio(RR), Pairs Completeness(PC), Pairs Quality(PQ) and F-score. The integrating results denote that each data source can get much Complementary Completeness(CC) and Increased Completeness(IC).