This PDF file contains the front matter associated with SPIE Proceedings Volume 13166, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

By utilizing the data from the infrared and water vapor channels of the FY-4A satellite, as well as the reanalysis data from the European Centre, we conducted a study on the identification method of convective cloud clusters during three different severe thunderstorm wind events. Based on the characteristics of convective cloud clusters observed by the satellite, we selected cloud cluster feature parameters that have significant early warning implications. We then performed parameter calculations and feature analysis on the identified convective cloud clusters. The conclusion is as follows: The multichannel threshold method has been improved by adding a parameter for the tropopause temperature. This method allows for a clearer observation of the uneven development of convective clouds. It can also clearly identify the center of strong updrafts within convective clouds. As the convective clouds rapidly develop, the infrared brightness temperature decreases rapidly, with a minimum brightness temperature below 220K. The convective clouds begin to intensify. Intense updrafts penetrate the tropopause, with cloud top temperature significantly lower than the tropopause temperature. The maximum brightness temperature gradient within the cloud is located at the leading edge of cloud clusters.

Semantic segmentation of remote sensing images is a critical task in computer vision, yet it has often been overlooked in the context of the images themselves. Given the high similarity between segmentation targets and the background in satellite remote sensing images, conventional deep networks tend to lose vital boundary features and contextual information, which are pivotal for accurate segmentation. To address this issue, I enhance the decoupled network architecture proposed by my predecessors. The improved network, named SplitNet, retrieves edge feature information from a shallow network and global features from the deep network applied to downsampled images. In a novel approach, I introduce a feature map fusion method that integrates edge, body, and global features, sharpening the network's focus on segmenting edge location features of the target. Our experiments demonstrate that SplitNet achieves substantial results on the DeepGlobe land classification dataset.

By the use of the Fengyun-4A geostationary satellite's multi-channel brightness temperature data (FY-4A). Examined and contrasted the physical properties of the cloud tops for three distinct thunderstorm gust episodes, each with varying intensities and ranges. The following outcomes from the analysis: Updrafts in convective clouds must achieve a specific range and strength. Positive BTD1(the difference between the infrared and water vapor channels) scores are also found in regions with wind speeds higher than ten levels. These regions experience freezing, a thickening of the cloud top, and a brightness temperature gradient greater than 20K. These regions are generally seen in front of the direction that clouds are moving. By using criteria such as BTD1 greater than 0 K, cloud top brightness temperature gradient greater than 20K can made preliminary judgments on the convective areas that could produce hazardous weather.

Aiming at the uncertainty of high slope deformation data and the problem that the actual deformation trend is easy to be overwhelmed by strong noise, this paper proposes an algorithm based on the coupling of set-averaged empirical modal decomposition (MEEMD) and singular value decomposition (SVD) to realize the noise cancellation of deformation data, firstly, the low-frequency flickering noise is rejected from the deformation data by using MEEMD decomposition to realize the primary filtering of the signal; and then, the SVD decomposition is performed on the obtained IMF components separately to reject the high-frequency white noise of each IMF component. Then, the SVD decomposition of each IMF component is carried out to eliminate the high-frequency white noise of each IMF component to realize the secondary filtering of the signal. The results show that the model has significant noise reduction effect and can fully describe the intrinsic detail information of the deformation data, which provides a fruitful method for future research on the processing of deformation monitoring data.

Due to rapid economic growth, urbanization, and population expansion, human society is facing increasing pressure on natural resources and the environment. Vegetation cover is closely related to these issues, and as a key indicator of ecological health, it is crucial for maintaining biodiversity, preventing soil erosion, and promoting economic and social well-being. Remote sensing technology, due to its high efficiency and low cost, has become the primary means of monitoring vegetation coverage. In this study, the Songhua Lake Basin in Jilin Province was selected as the research area. Landsat TM/ETM+ data and Normalized Difference Vegetation Index (NDVI) were used to calculate and analyze the vegetation coverage in the Songhua Lake Basin in 2000 and 2006, employing the dimidiate pixel model. The obtained vegetation coverage data provide a scientific basis for targeted conservation measures.

Cross-view geolocation, which aims to geolocate ground-view images using reference satellite imagery, is a challenging task that requires effective strategies to minimize significant disparities between images. In this paper, we propose a novel approach to address this challenge. By exploiting the projection relationship between ground images and satellite images, we are able to convert ground-view images into satellite-view images, thereby mitigating the inherent disparities between the two perspectives. To enhance the converted images further, we employ a conditional generative adversarial network (CGAN). This network generates satellite perspective images that exhibit greater consistency with the actual satellite data, thereby reducing excessive disparities and improving overall image quality. Additionally, we adopt a joint training approach in a multi-task setting, wherein we synthesize satellite images from ground images and conduct cross-view image matching. This framework facilitates mutual learning between the tasks and leads to improved performance. Empirical evaluation of our proposed method showcases significant advancements in terms of retrieval accuracy and synthesis quality when compared to existing techniques. These findings underscore the potential of our approach in addressing the challenges associated with cross-view geolocation.

Taking the Yishui County, Shandong Province as the research area, we processed GF-2 images from different periods in May 2018 and May 2020. Based on their spectral and texture differences, combined with topographic and geomorphic features, multiple hidden geological hazard points are identified. Finally, we conducted on-site investigations and ultimately identified multiple potential geological hazards.

Geometric-correction is used to correct the distortion between image and corresponding ground target, it is one of the most time-consuming steps in remote sensing data processing, thus becoming a bottleneck restricting the effectiveness of remote sensing data. Therefore, a fast geometric-correction method of remote sensing image, based on GPU parallel processing, is proposed in this paper. First, a three-dimensional (3D) linear transformation is introduced to reduce the computational amount of coordinate projection conversion. Then, the GPU parallel mapping is conducted. Furthermore, two performance optimization methods are used to further improve the efficiency. The approach is tested with GF-1 panchromatic image with Intel i7 CPU and NVIDIA Tesla M2070 GPU. The experimental results show that the processing time is only 2.08 s, with a speedup ratio of 142.42, which could meet the requirement of fast geometric-correction for big-data remote sensing images.

In response to the problem that traditional surveying and mapping methods are difficult to adapt to complex terrain environments and have low accuracy in extracting geographic information spatial features, a research on geographic information spatial feature extraction based on drone aerial surveying and mapping technology is proposed. Based on unmanned aerial vehicle (UAV) aerial surveying and mapping technology, geographic information spatial images are obtained, and offset parameters and distortion coefficients are introduced to achieve the acquisition and correction of geographic information spatial images; Cluster the geographic information spatial features of similar targets into a set of set data, and convert their coordinates using world coordinates and pixel coordinates as references. Using orthogonal vectors between geographic information spaces to describe geographic information spatial features and achieve geographic information spatial feature extraction. The experimental results show that the highest elevation accuracy of the geographic information spatial feature extraction method in this article is 0.5m, with an average data coverage of 97.5% and an average data consistency of 97.5%.

Apart from being essential to the development of grasslands and crops, soil moisture has a significant impact on the water cycle and the global climate. Drawing from multispectral Landsat 8 OLI photos, Added to this study, soil moisture was simultaneously measured in the field based on topographic parameters and surface biophysical features. Soil moisture inversion models of multivariable linear regression (MLR) and random forest (RF) were constructed by empirical model method. And compared with inverse distance weighting (IDW). The findings demonstrated the dependence of surface soil moisture on surface biophysical characteristics was high, with the highest inversion accuracy of random forest, in which the R² of RF model was above 0.8 in both months, which was at least 8% higher than that of MLR model, and at least 11% higher than that of IDW model, and the corresponding RMSE was the smallest in both months. In various months, the main variables influencing the soil moisture content were elevation and surface temperature. The investigation's area had low overall soil moisture content, with most of it having less than 20. High, steep slopes in the north and southeast of the area had higher soil moisture content, whereas the region's flat areas in the north-central region had lower soil moisture content.

Urban land use extraction is an important method to accurately obtain land use information. The object-oriented classification method is mainly used to extract and analyze the land use information of Anshan urban area spanning 10 years from 2009 to 2014 by using spectrum, texture information and vegetation index. The results show that in 2009, 2014 and 2019, the overall classification accuracy was 91%, 87% and 86%, and the kappa coefficient was 85.92%, 79.86% and 77.41%, with good classification accuracy; This paper analyzes the changes in time and space of land use in Anshan City from 2009 to 2019, indicating that the urbanization process of Anshan City has been relatively rapid in the past decade.

In lunar science, volatiles are defined as chemical elements and unstable compounds, evaporating, sublimating or flowing in other ways at low temperature. At moderate temperature, vapor transfers its flowing elements from solid materials to coexisting gas phases. Such substances include S, Cu, Zn, As, Se, Ag, Cd, In, Te, Hg, Tl, Pb, Bi and halogens (F, Cl, Br, I). These volatile substances are mainly added by small impactors, so they should be uniformly deposited in lunar regolith. In the PSRs of the lunar polar region, the extremely low temperature may allow the solar activity record to be kept longer. When the lunar soil is heated, it will release volatile elements in the form of gas, especially the particles injected into the lunar soil by the solar wind. In deeper lunar soil, it means that these particles were implanted in different periods of solar evolution. The study of regolith in polar PSRs may help to solve this problem, because it has not been widely heated and therefore may not be subjected to post-implantation modification. The research on volatiles detection in lunar polar region can help us confirm the content and distribution of water ice in the permanent shadow region of the lunar polar region, and the composition, distribution, source, transport and loss mechanism of other volatiles in this region, as well as the relationship between lunar soil in the polar region and the ancient solar wind environment, and the environment and mineral resources in the region.Chandrayaan-3 provided significant assistance to the conceptualization of this project. At the end of the article, we discussed the lunar orbit.

Mindoro Island is located in south of Luzon (Philippines). It is a complex junction and considered an emplaced Cenozoic South China Sea oceanic lithosphere as a result of the collision between the Palawan microcontinental block and the Philippine. In this paper, high resolution satellite images have been used to derive lineament maps by geological interpretation. By utilizing satellite imagery, various remote sensing techniques, and image processing algorithms, researchers and geologists can identify and characterize fault structures, gaining insights into the tectonic activity and geological processes in the area. In addition, it is feasible to explore the active faults in weak active structural area by using satellite remote sensing technique. The results show the main direction of the structure line is consistent with the main island topography and ridge direction. The fault structure is mainly NW trending in the north. In the south of the center, fault structure becomes complicated.

Dongting Lake, located in Hunan Province, China, is the second-largest freshwater lake in the country and a crucial ecosystem and biodiversity hotspot. Investigating the temporal and spatial evolution of Dongting Lake's water areas is conducive to regional water resource development and provides assurance for regional ecological balance. In this study, Landsat-8 OLI remote sensing images from the years 2017 to 2021 were utilized. The normalized difference water index (NDWI) water threshold value method was employed to extract the water area of Dongting Lake. The study analyzed the temporal and spatial evolution of Dongting Lake, delving into the migration direction of the Dongting Lake basin's centroids, providing a reference basis for the sustainable development of Dongting Lake. The results indicated that the water area of Dongting Lake decreased from 913.65 km² in 2017 to 723.14 km² in 2019, followed by an increase to 866.65 km² in 2021. The centroids of Dongting Lake shifted southwestward from 2017 to 2019 and then east-southeastward from 2019 to 2021.

Domestic GF-2 sub-meter level satellite images have revealed a great application potential in geological exploration. However, the satellite images have been little documented in lithology and structure identification. Taking the Heijinshan area of Dunhuang City in Gansu Province as an example, this work established the technical flowchart of the GF-2 data, preprocessed the GF-2 four-band fused reflectance data by radiometric calibration and atmospheric correction, and conducted geological interpretation using visual interpretation methods. Quartz veins in the Heijinshan area were also extracted based on the decision tree algorithm. This study revised previous geological mapping, delineated precise geological and lithological boundaries, and newly identified the distribution of two basalt sections within the lower member of the Lower Carboniferous Hongliuyuan Formation. A five-layer decision tree was constructed to eliminate interfering features layer by layer, and more than ten large-scale quartz veins were well identified. These quartz veins were mainly distributed in the metamorphic clastic rocks of the lower Hongliuyuan Formation of the Lower Carboniferous, and were also developed in the Late Permian diorite granite. This study indicates that GF-2 remote sensing data can play a dominant role in the interpretation of lithology-structure and quartz veins.

In recent years, with the advancement of remote sensing technology, there has been an increasing reliance on computer graphics to render terrain targets and obtain intuitive images. However, the lighting properties of material surfaces are crucial physical factors that impact the rendering effect, directly influencing the visual appearance of the target under various lighting conditions. This article delves into the diversity of materials and the intricacies of light reflection, examining the modeling methods and simulation effects of material surface lighting properties. By leveraging the intersection test results of objects and light, we employ the Blinn-Phong lighting model to introduce the optical characteristics of remote sensing target micro surface materials for simulation. This lays the foundation for remote sensing image stitching and other related tasks.

This study is selected central urban areas(CUA) of Hunan Province as the research object.The land use data and socioeconomic data were used in this study to analyze the evolution characteristics and its drving mechanism of the CUA of Hunan province during 1990-2020 at the macro and micro levels.The research methodology includes the development of the spatial database, construction of the grid system, Geographically and temporally weighted regression (GTWR) model and building of the peak map model PMM).Analyze the coupling relationship between expansion characteristics and core driving factors based on the actual development and land use situation of each CUA. The main conclusions are as follows: (1) From 1990 to 2020, urban land area in the central areas of Hunan Province showed an increasing trend, with high-speed expansion predominantly occurring in the eastern regions, and spatial development becoming increasingly uneven over time. (2) The expansion areas of the central areas of Hunan Province evolved from a centralized single-point distribution to a balanced distribution, undergoing three stages: "low-speed imbalance," "high-speed imbalance," and "medium-speed balance." The internal development coordination and overall coordination gradually improved. (3) From 1990 to 2010, the expansion pattern in the central areas primarily followed an "edge-remote" evolution pattern. However, after 2010, due to differences in the relationship between average urban expansion and cluster dispersion distance, expansion patterns began to diverge. The fact that clusters were too distant from the central core was an important reason for the loose spatial structure and slow development of central and western central areas. (4) The expansion of each central urban area is driven by various factors. Industrialization, policies, transportation, people's livelihoods, and cultural and educational factors being the main factors, but there are differences in their impact on the extent and morphology of expansion.

Based on residents' activity trajectory data to characterize urban vitality, using POI data and remote sensing image data to locate the spatial location of urban health resources, we study the reasons for the differences in the distribution of health resources in Hengyang city from the users' point of view and on the basis of multiple spatial characteristics. The study shows that: ① health resources are often distributed in the center of the city due to the influence of urban development history and planning policy, and there is a lack of resources in the peripheral areas of the city; ② health resource-intensive areas are often preferred by users, thus resulting in more resources in the center of the city and fewer resources in the peripheral areas; ③ compared with the center of the city, the peripheral resources have low accessibility, which is unfavorable for the arrangement of the resources and the performance of the function. The imbalance between the types of resources in the city center and the periphery due to geographic location affects the travel choices of the residents, which reveals the differences in the distribution of resources; finally, the corresponding resource allocation plan is formulated in response to the reasons for the differences in the distribution of health resources, with the intention of providing scientific references for the rational distribution of health resources.

In order to quantitatively identify the evolution of the runoff time and space of the basin, analyze the factors affecting the changes in the hydrological situation of the basin. This article uses the Western Han Dynasty Water Basin as the research area. Based on remote sensing technology, the impact of LUCC and NDVI changes on the runoff of the area. Firstly, the method of time and space evolution of the elements of various elements in the research area of the research area was analyzed by methods such as Mann-Kendall. Period (1990 to 2020) Vegetarian coverage (NDVI) transferred in space. The results show: (1) From 1960 to 2022, the runoff in the West Han River Basin showed an overall downward trend. From north to south, the runoff in the basin showed an increasing trend of 2.68×10⁸m³, 7.55×10⁸m³, and 12.99×10⁸m³ . The mutation years were concentrated in the 1980s and 1990s. (2) From 1980 to 2020, precipitation in the basin generally showed a slow upward trend in time, and spatially showed an increasing trend from upstream to downstream; temperature changes showed a consistent upward trend with the same increase in time, and spatially showed that the temperature in the upstream was lower than that in the downstream , consistent with global temperature changes. (3) The transformation of LUCC in the West Han River Basin from 1980 to 2020 was mainly due to the increase of forest and grass area and the decrease of cultivated area. (4) The NDVI in the West Han River Basin from 1990 to 2020 was at a medium to medium high level in the time series, and showed an overall upward trend. In terms of spatial distribution, the pattern of "downstream is better than upstream", and the vegetation cover was significantly improved, but contrary to the change of runoff, the increase of NDVI had a decreasing effect on runoff.

Marine reclamation took place at large scale in Lianyungang city of Jiangsu province, China during last two decades in order to provide extra land for economy growth. However, large land subsidence usually occurs on newly reclaimed land, which may bring damages to buildings and infrastructures constructed over the reclaimed land and seriously hinder the regional development in turn. Therefore, knowing detailed information about land subsidence over the reclaimed land is of great significance. As a new remote sensing monitoring method, synthetic aperture radar interferometry technology has great advantages in ground subsidence monitoring. Therefore, in this paper, the Sentinel-1A SAR images acquired from 2016 to 2021 are processed using InSAR time series technology to monitor and analyze the spatio-temporal characteristics of ground subsidence in the region. The results show that by the end of 2021, large subsidence occurred over the reclamation area of Lianyungang which is mainly caused by soil consolidation. The maximum subsidence rate is -165 mm/year, and the maximum cumulative subsidence reaches -991 mm during the 6 years, which is located in the northeast of Lianyungang. The deformation monitoring of the region has certain theoretical references for disaster prevention and control.

The prolonged and prolonged spread of the global epidemic has disrupted people's lives, affecting every organization and every person in the world. Our daily life and work patterns have changed completely, and with them, there are new needs that have not yet been met. Office building space is one of the areas most seriously and profoundly affected by the COVID-19 epidemic. In order to adapt to the changes in the post-COVID-19 era, the components and design concepts of office building space will inevitably evolve and change continuously. This paper conducts research and case analysis on new office building space in the post-epidemic era, and forms the design strategies and methods of new office building space in the post-epidemic era to provide effective reference and research significance for the occurrence of similar types of public viruses in the future.

Based on the current quality detection methods and development direction of digital chart, a Chart quality detection method based on depth range constraints in isobath zones was proposed for a large number of water depth points and isobates on digital chart. Firstly, in view of the defects existing in the current quality detection method, the current different quality detection methods are compared and analyzed. Then, the logical error of the chart data is studied. Through experimental calculation, the weight of the depth range extraction is determined. The experimental results show that the proposed method can effectively improve the accuracy of logical detection.

In recent years, frequent sea ice disasters have led to severe disruptions in marine ecology and hindered maritime transportation. In this study, Sentinel-1 data from 2016 to 2022 were utilized to extract texture features from images. A comparative analysis of the classification accuracy of three models—Support Vector Machine, Iterative Self-Organizing Clustering, and Random Forest—was conducted. A high-precision sea ice classification model was established to analyze spatiotemporal changes. The results indicate that the Support Vector Machine model exhibited the highest accuracy, with an overall accuracy of 87.61% and a kappa coefficient of 81.42%, demonstrating the model's stability.

The smoothing process of the roaming path coordinates and camera direction at each path point proceeds with the Gaussian filtering algorithm. The resultant connected roaming path exhibits enhanced naturalness and smoothness, mitigating the abruptness between path points. The processed virtual camera, during movement, demonstrates smoother and more fluid changes in direction, thus avoiding sudden camera shifts and reducing both visual jitter and user discomfort. Employing multithreading techniques, a dedicated timer thread ensures the consistent update of the roaming virtual camera's position and direction at a fixed frame rate. This approach diminishes the incongruent effects of virtual camera motion caused by varying frame rates and addresses inconsistencies in user input response times. The proposal enhances the temporal consistency, predictability, and physical simulation stability of the virtual camera's runtime performance. A quantitative assessment of the smoothness of the camera's turning along the path is introduced by calculating the root mean square (RMS) value of camera direction changes. This methodology is applicable for detecting abnormal paths, optimizing paths, and comparing the smoothness of camera turns along distinct routes.

Hierarchical Temporal Memory (HTM), an unsupervised online algorithm, has been validated in multiple fields of machine learning, especially prediction and anomaly detection, which can well reduce the false alarm rate by analyzing from multiple dimensions. For instance, traffic flow prediction, ocean observation data anomaly detection, etc. However, the analysis and prediction based on the real-time monitoring data of technical engineering have not been addressed. In this study, a real-time prediction method of foundation pit deformation based on HTM is proposed to monitor the sensor data of different depths in real time, predict the horizontal displacement and vertical settlement, set the abnormal range according to the requirements of the monitoring accuracy level, and calculate the measured and predicted values. Combining the measured and predicted values by cloud computation, a signal value is returned to determine whether it is anomalous or not. The error reporting rate is set according to the number of anomalies. The results show that the HTM can be used to analyze and predict the pit deformation monitoring data effectively and reliably.

To enhance the accuracy and efficiency of detecting verticality in single-pile construction for offshore wind power, a binocular stereo vision measurement system is developed. This system combines binocular stereo imaging and attitude measurement to enable rapid and precise measurement of pile verticality in dynamic environments. The verticality of the inspected pile is determined using a total station, and the results are compared with the verticality data obtained from the binocular stereo vision system. Simulation experiments reveal that when the total station detects a verticality of 2.96‰ for the cylinder, the average verticality detected by the binocular stereo vision system in dynamic environments is 3.05‰. The minimum difference between the detection results of the binocular stereo vision system and the total station is 0.5‰, while the maximum difference is 1.27‰. The binocular stereo vision system exhibits high measurement accuracy, stability, and real-time performance, thereby enriching measurement methods and improving work efficiency in actual construction processes.

This study proposes an intelligent data update method based on ArcGIS Server to address the problem of untimely and inaccurate updates of surveying and mapping geographic information data. This method achieves data updating through the following steps: firstly, obtaining surveying and mapping geographic information data through various methods, and integrating and visualizing it using the map service function of ArcGIS Server; Secondly, data preprocessing techniques are adopted, including coordinate conversion, format conversion, data cleaning, etc., to improve the accuracy and readability of the data; Then, using the editing function of ArcGIS Server and spatiotemporal database technology, the surveying and mapping geographic information data is updated in real time and stored in the database. The experimental results indicate that this method can effectively improve the efficiency and accuracy of updating surveying and mapping geographic information data.

Timely and accurate agricultural monitoring can provide effective guidance for crop management but the process can be challenging. In this study, a GF-2 remote sensing image was used as the research object to extract the winter wheat based on an improved deforming full convolutional neural network (DFCN) model. The deformable convolution model was developed to try to better extract geometric features and object-level semantic information and improve the ability to describe the different spatial distribution characteristics of winter wheat. Convolution deformation was obtained by adding a trainable two-dimensional migration to each convolution layer in the network. The DFCN network model was compared to traditional neural networks such as FCN and U-Net. The precision ratio of DFCN was improved to 98.1%,and the time required was reduced to 0.63 seconds. In contrast, the accuracies of winter wheat automatic interpretation obtained by FCN and U-Net were 89.3% and 93.9%, respectively. The experimental results show that using the DFCN model to interpret winter wheat automatically can achieve the highest accuracy, better capturing its morphology and spatial distribution, and with good generalization performance. Through the training of the input data set and parameter tuning, a high-precision intelligent and DFCN model with high practicability and reliability was constructed, and a high-precision automatic remote sensing identification system for typical crops in the north of the Huang-Huai-Hai Plain was realized.

With the gradual improvement of China's infrastructure construction level, the importance of ensuring railway safety is increasing. The traditional railway limit measurement has the disadvantages of no data retention and difficult operation. How to obtain measurement data efficiently and accurately has become an urgent problem to be solved. This paper proposes to apply the three-dimensional mobile laser scanning technology to the limit measurement, and carry the technology on the railway measurement robot, and compare the measurement results with the traditional total station. It has been proved that the railway measurement robot has the advantages of high degree of automation, fast data acquisition speed, large amount of point cloud data, high accuracy and reliability, and rapid data processing. In the future, this technology can be promoted and applied in similar engineering measurement projects.

In the closed-loop fiber positioning control mode of Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST), there are stringent requirements for time efficiency. However, due to the large size and high resolution of the images required for fiber positioning, the data volume for a single image is extremely large, reaching up to 90 MegaBytes per image. The current time efficiency of the closed-loop control system has yet to meet the requirements. This paper attempts to implement the fiber positioning algorithm, the light centroid algorithm, for calculating fiber positions in an FPGA-based edge fiber positioner detection module. Optimizing this module can significantly save image transmission time. The article mainly utilized Vivado HLS to port the algorithm to an FPGA and achieved improvements to its implementation. To enhance the algorithm's computational efficiency, the labels were assigned to each light spot to calculate the centroid coordinate, which enabled more efficient operation on the FPGA. The results indicated that the new hardware architecture effectively have reduced transmission time, making the improved centroid algorithm more suitable for operation on an FPGA.

This research employs CiteSpace software to analyze research outcomes on Abnormal Behavior Detection Based on Video Surveillance. It focuses on the China National Knowledge Infrastructure Database (CNKI) and the "Web of Science" Database, aiming to identify research trends, provide some references for advancing this research direction. The following conclusions are drawn: (1) The literature on abnormal behavior detection models and algorithms has shown a consistent increase over time. (2) The current research trend is deep learning, video surveillance and abnormal detection. Research tends to diversify. (3) In China and abroad, there is more attention in the field of deep learning, video surveillance and anomaly detection.

Due to the high popularity of ice and snow tourism in Harbin at the end of 2023, the passenger throughput of the airport has increased significantly. In order to improve the efficiency of security inspection, Harbin Taiping Airport has introduced a face recognition system. However, due to the large number of passengers, the current airport face recognition system can’t accurately detect the passenger's face part when the number of people increases. In order to improve the speed of the face recognition system and reduce the errors caused by the increase of the number of faces in one frame, this paper proposes a new face detection technology and method to identify the identity of airport passengers. Combining the characteristics of the airport recognition system that requires fast response speed and high detection efficiency, this paper focuses on discussing the face detection algorithm. The security subsystem based on face detection is also studied. This paper describes the completion process of the airport face detection and recognition system: face classifier cascade, the airport face detection, the feature extraction and feature recognition, the final judgment of personal information after recognition, and discusses the method of graphically completing the whole system, and proves the robustness of its face recognition.