Secondary organic aerosol (SOA) is an important component of organic aerosol (OA), yet its atmospheric evolution and impacts on volatility remain poorly understood. In this study, we investigated the volatility of different types of SOA at a downwind site of the Pearl River Delta (PRD) region in the fall of 2019, using a time-of-flight chemical ionization mass spectrometer coupled with a Filter Inlet for Gases and Aerosol (FIGAERO-CIMS). Positive matrix factorization (PMF) analysis was performed on the thermogram data of organic compounds (referred as FIGAERO-OA) measured by the FIGAERO-CIMS. Eight factors were resolved, including six daytime chemistry related factors, a biomass burning related factor (BB-LVOA, 10 % of the FIGAERO-OA), and a nighttime chemistry related factor (Night-LVOA, 15 %) along with their corresponding volatility. Day-HNOx-LVOA (12 %) and Day-LNOx-LVOA (11 %) were mainly formed through gas-particle partitioning. Increasing NOx levels mainly affected SOA formation through gas-particle partitioning, suppressing the formation of low-volatile organic vapors, and thus promoting the formation of relatively high volatile OA with a higher N : C ratio. Two aged OA factors, Day-aged-LVOA (16 %) and Day-aged-ELVOA (11 %), were attributed to daytime photochemical aging of pre-existing OA. In addition, the daytime formation of Day-urban-LVOA (16 %) and Day-urban-ELVOA (7 %) could only observed in the urban plume. Results show that both gas-particle partitioning (36 %) and photochemical aging (30 %) accounted for a major fraction in FIGAERO-OA in the afternoon during the urban air masses period, especially for high-NOx-like pathway (similar to 21 %). In general, the six daytime OA factors collectively explain the majority (82 %) of daytime SOA identified by an aerosol mass spectrometer (AMS). While BB-LVOA and Night-LVOA accounted for 13 % of biomass burning OA and 48 % of nighttime chemistry OA observed by AMS, respectively. Our PMF analysis also demonstrated that the highly oxygenated OA and hydrocarbon-like OA cannot be identified with FIGAERO-CIMS in this study. In summary, our results show that the volatility of OA is strongly governed by its formation pathways and subsequent atmospheric aging processes.

Post-disaster structural safety assessment requires robust and intelligent techniques capable of integrating heterogeneous sources of information. This paper presents a novel parallel interactive fusion framework that combines two-dimensional (2D) image segmentation with three-dimensional (3D) point cloud analysis for post-fire damage assessment of reinforced concrete components. Unlike conventional approaches that treat visual and geometric features independently, the proposed method establishes a bi-directional interaction mechanism, in which damage-sensitive regions extracted from2D segmentation guide the sampling and weighting of 3D features, while geometric context from point clouds refines pixel-level recognition. A dual-branch deep network is designed, incorporatingYOLOv10-SimAM-based 2D feature extraction and a 3D fusion transformer module for heterogeneous feature alignment. Extensive experiments on a curated dataset of fire-damaged concrete images and reconstructed point clouds demonstrate that the proposed method achieves significant improvements in accuracy and robustness compared with state-of-the-art vision-only and geometry-only baselines. Furthermore, the framework enables precise localization and quantification of multiple damage types under highly variable fire scenarios, addressing the inherent challenges of incomplete, noisy, and multi-scale data. The study illustrates the potential of multi-source fusion learning in enhancing situation awareness for structural resilience and provides a promising paradigm for applying information fusion in safety-critical civil engineering infrastructures.

Semantic segmentation is a critical task in remote sensing applications. With the continuous increase in the diversity of available remote sensing data, semantic segmentation has gradually evolved from single-source image interpretation to multisource feature fusion. However, many existing multisource methods extract features from each input source and then combine them through simple channelwise concatenation or addition. Although this strategy can introduce complementary information, it may also produce redundant or noisy representations, especially when the input sources are highly correlated, such as multi-angle polarization images or optical-elevation data. To address this problem, we propose a multisource semantic segmentation model named DSMF-Net (Dual Stem Multisource Fusion Net), which adopts an encoder-decoder architecture and introduces two dedicated fusion modules: a convolutional neural network (CNN) stem and a Transformer stem. The CNN stem focuses on shallow spatial-detail fusion, whereas the Transformer stem models global dependencies among deeper multisource features. In addition, a multidilated convolution decoder (MDCD) is used to restore multiscale spatial details. Experiments on the Potsdam, Vaihingen, and ZJU-RGB-P datasets show that DSMF-Net achieves competitive segmentation accuracy in both optical-elevation and polarization-based multisource scenarios.

As the spatial resolution of remote sensing imagery continues to improve, Earth observation scenes exhibit greater diversity in spatial and spectral characteristics, together with increasingly complex surface structures across multiple scales. In such scenarios, global land-cover patterns and local high-frequency details are often strongly coupled, posing significant challenges for accurate scene interpretation and boundary delineation. From a feature representation perspective, spatial-domain features are effective for modeling global context and high-level semantics, whereas frequency-domain representations facilitate the separation of structural components at different scales and enhance fine boundary details, but have limited capability in explicitly capturing semantic relationships. Consequently, most existing semantic segmentation methods operate in a single domain, either spatial or frequency, which restricts their ability to jointly model semantic consistency and diverse spatial structures. To address this issue, we propose a semantic segmentation network termed Spatial–Frequency Collaborative Modeling Network (SFMNet), which jointly learns complementary representations in both spatial and frequency domains. By leveraging spatial-domain semantic priors to guide frequency-domain structural enhancement, SFMNet enables coordinated modeling of land-cover semantics and multi-scale surface structures. In addition, cross-scale interaction between low- and high-frequency components and adaptive multi-level feature fusion are introduced to improve structural consistency across hierarchical feature representations. Extensive experiments on the ISPRS Potsdam, Vaihingen, and LoveDA datasets demonstrate that SFMNet consistently outperforms existing methods in both overall segmentation accuracy and boundary delineation quality. Specifically, SFMNet achieves mean Intersection-over-Union scores of 81.30%, 70.23%, and 55.07%, and Boundary mean Intersection-over-Union scores of 65.42%, 56.76%, and 43.74% on the three datasets, respectively.

Despite numerous field studies focusing on the chemical composition of particulate matter, systematic investigations on particulate chloride (Cl − ) are scarce in China. This study examines the spatiotemporal distribution, origin, and environmental impact of Cl − across China via compiling a comprehensive data set of submicron aerosol (PM 1 ) detected by the aerosol mass spectrometer (AMS). In addition, we integrated high‐time resolution aerosol composition data from AMS at six representative sites in China for more detailed insights. The mass concentration of Cl − across China is notably high (1.9 ± 2.7 μg m −3 ) in comparison to the global average (0.3 ± 1.0 μg m −3 ), with a distinguished N–S pattern (North > South) and a distinct seasonality (winter > other seasons). Three categories of Cl − diurnal variations were classified, suggesting the primary combustion emission and gas/particle partitioning are the main drivers for the dynamic variation of Cl − . The good correlation between Cl − and combustion tracers corroborates the assertion that coal and biomass combustion are the main anthropogenic sources of Cl − based on emission inventory. Furthermore, the quantified environmental impacts of Cl − on ammonium balances, aerosol liquid water content, and hygroscopicity were systematically explored. In extreme cases, the Cl − can enhance 100% ALWC during polluted periods, signifying its key role in impacting the physiochemistry of aerosols. The mutual promotion between Cl − and other environment effects was also found. In summary, this study enhances our understanding of the distribution, sources, and environmental effects of Cl − across China, indicating that Cl − should be systematically considered in elucidating environment effect of fine particles.

This study focuses on the on-orbit attenuation tracking problem of FY-3B/MERSI.Ocean surface sun glints are adopted as stable targets to track the attenuation from 2011 to 2018.Based on cloud-free and effective sun glint area data,the 865 nm band is regarded as the benchmark band;the ratios between other bands' reflectance and the benchmark band's reflectance are calculated to analyze the attenuations of these bands during the 8 years.There are obvious degradations for all FY-3B/MERSI bands,especially for shortwave bands.The annual degradation rate of 412 nm is 7.12％,while the corresponding value is 0.28％for the 765 nm band.The degradation is much bigger for the 1 030 nm band,at around 3.88％.Furthermore,the reflectance ratios between different bands and benchmark bands show obvious oscillation,consistent with the north-south periodic change of latitude at the center of the sun glint.Ocean surface sun glints are an effective target for the inter-band radiometric calibrations,and could help track the long-term attenuation of on-orbit sensors.

Particulate levoglucosan is an important tracer for biomass burning emission in ambient air. However, recent studies question its reliability as a biomass burning tracer in Chinese mega cities due to important contribution from potential non-biomass burning sources, such as cooking. To address this, we examined the dynamic variation and sources of levoglucosan using a chemical ionization mass spectrometer and other advanced instruments during Beijing's summer of 2021. The average mass concentration of levoglucosan and its isomer (C6H10O5) was 0.025 +/- 0.014 pg/m(3) , constituting 0.55 % +/- 0.32 % of total organic carbon (OC) in this campaign. Despite cooking emissions contributing significantly to the organic aerosol (OA, 20 %), levoglucosan and its isomers correlated more strongly with biomass-burning related tracers ( R > 0.6), black carbon (R = 0.72) and less so with cooking-related sources (R = 0.3). This indicates that levoglucosan is primarily dominated by biomass-burning emis- sions rather than cooking in Beijing's urban areas during summertime. The diurnal varia- tion of levoglucosan concentrations highlighted the importance of daytime and nocturnal biomass burning emissions during polluted periods in Beijing. Using levoglucosan as a tracer to quantify the biomass burning OC (BBOC), we found good agreement on the time series of BBOC between the tracer method and other independent source apportionment method. This reaffirms the reliability of levoglucosan as a biomass burning tracer. Biomass burn- ing contributed an average of 7 %-8 % to OC, highlighting its significant impact on Beijing's summer air quality. Our study enhances understanding of biomass burning influences on ambient aerosol in typical urban areas. (c) 2025 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences. Published by Elsevier B.V.

On September 7, 2021, the GaoFen5-02 (GF5-02) Satellite, the new generation of Chinese hyperspectral remote sensing satellite was successfully launched. GF5-02, the successor to the GF5 satellite, was equipped with six advanced hyperspectral payloads. One of the most important payloads onboard the GF5-02 satellite, the advanced hyperspectral imager (AHSI) has a spatial resolution of 30 m, 330 bands in a spectral range of 380-2500 nm. The spectral resolution of the visible and near infrared (VNIR) and short-wave infrared (SWIR) bands are better than 5 and 10 nm, respectively. To analyze the spectral performance of the GF5-02 AHSI, an on-orbit spectral calibration method that utilizes atmospheric limb observations with an on-board calibration system was proposed in this letter. The on-orbit spectral calibration results were validated by atmospheric absorption features with synchronous measurements of surface reflectance and atmospheric parameters. For the GF5-02 AHSI, the shifts in the central wavelength of the VNIR band is 0.117nm, while the shifts in the full width at half maximum (FWHM) is 0.02 nm. In the SWIR band, these values are 0.25 nm for the central wavelength and 0.04 nm for the FWHM. The results demonstrate that the applied method is effective for on-orbit spectral calibration for GF5-02 AHSI.

Rapid urbanization in China has exacerbated the dual challenges of urban heat islands (UHIs) and air pollution, threatening urban sustainability. We conducted a national-scale analysis of the spatiotemporal dynamics and synergy between the surface UHI intensity, distinguished as daytime (DUHI) and nighttime (NUHI), and major air pollutants (PM2.5, PM10, NO2) in 370 Chinese cities (2000–2019). Using multi-source remote sensing, ground-based monitoring, and urban data, we applied coupling coordination and correlation analyses to quantify these interactions. Key findings reveal distinct patterns: (1) The annual mean land surface temperature (LST) rose, with the nighttime LST (NLST) increasing faster than the daytime LST (DLST). Conversely, the UHI intensity showed an overall decline, with the DUHI decreasing more than the NUHI. (2) Air pollutants displayed strong seasonality; while PM10 concentrations decreased slightly over the long term, NO2 levels rose significantly. (3) Monthly, pollutants correlated negatively with LST (R2 > 0.92 for PM2.5), suppressing the DUHI but intensifying the NUHI. Long-term, the correlation trend revealed a strengthening synergy, particularly between particulate matter and NUHI (trend R2 = 0.50). (4) Spatially, over 90% of cities exhibited high UHI–particle coordination. Key associated factors include anthropogenic activities, urban morphology, and natural mitigation factors. We conclude that disrupting the heat–pollution synergy requires integrated strategies, namely reducing emissions at the source, optimizing the urban form, and enhancing ecological regulation. This is essential for advancing low-carbon, climate-resilient urban development.

Reactive organic carbon (ROC) is the sum of both gas- and particle-phase organic compounds excluding methane, serving as the fuel for atmospheric oxidation processes. Comprehensive characterization of organic mixtures, however, has been a long-standing challenge. Here, we investigate the speciation, properties, and evolution of ROC in a holistic view based on comprehensive field observations using four advanced mass spectrometers at an urban site in Guangzhou, a megacity in southern China. The summed concentration of over 1000 organic species detected in the gas phase averaged 124.7 μgC/m3, with oxygenated organics accounting for the largest fraction (44% by carbon mass, the same below), whereas the concentration of organic aerosol was 8.1 μgC/m3 on average. The observed ROC was dominated by volatile species (84%), while semi- and intermediate-volatile species that were not routinely measured contributed 10%. C1-C8 compounds constituted the major fraction of ROC (85%), most of which were long-lived oxidation products, along with anthropogenic alkanes and aromatics. Over 16 h of photochemical aging, the observed ROC mass decreased by 14%, which was much lower than in previous attempts in urban air, highlighting a better carbon closure with the advancement of mass spectrometry techniques. Our work provides insights into the evolution of the speciation and properties of ROC during oxidation processing.

For resource-based cities, the rapid development of industrialization and urbanization has led to significant carbon emissions (CEs), accelerated the rise of urban land surface temperatures (LSTs) and hindered sustainable urban development. This study constructed a model to measure the carbon-heat relationship to clarify the complex relationship between LSTs and CEs in resource-based cities. The results show that:1) High-temperature areas are primarily concentrated around the urban center and large industrial zones, with average LSTs reaching a peak of 35.7 degrees C in 2015, indicating severe temperature polarization; 2) CEs exhibited an overall upward trend with a diffusion effect, particularly pronounced in the urban center and industrial zones. Areas with extremely significant, strong significant, and generally significant growth in CEs accounted for 4.64%, 3.81%, and 81.35%, respectively, showing a concentrated increase in the urban center; 3) A positive correlation between CEs and LSTs of the city was identified, and the distribution of urban heat island and the high value area of CEs are concentrated and similar; 4) The synergistic effects between LSTs and CEs varied between urban center, suburban and peripheral areas, due to human activities. Areas with a high positive correlation between CEs and LSTs are concentrated in urban centers and peripheral areas, while for urban suburbs, the correlation is weak or even absent. To mitigate the negative effects of carbon-heat accumulation, urban centers should avoid high population concentrations, and the carbon sink potential of green spaces near industrial zones and peripheral areas should be fully utilized. These insights provide actionable strategies for sustainability of resource-based cities, particularly in the governance of urban thermal environments and the mitigation of CEs.

Semantic segmentation of remote sensing images is crucial for resource exploration, precision agriculture, and environmental monitoring. However, conducting semantic segmentation on single-modality data for remote sensing images that contain various scenes, especially unique scenes, is highly challenging. To address this challenge, we propose SiMultiF, a Siamese architecture-based multimodal feature adaptive fusion semantic segmentation network. SiMultiF employs a dual-branch Siamese structure feature extractor. The adaptive feature weight adjustment module (AFWAM) and the multimodal fusion module (MFM) facilitate in-depth understanding and extraction of multimodal data. Specifically, the Siamese structure can extract features from multimodal data concurrently without adding to the number of parameters. The AFWAM module can adaptively identify the importance of different modal data and dynamically adjust the modal weight to enhance the network's comprehension of complex scene data. Additionally, the cross-attention (CA)-based MFM module bridges modality gaps and achieves comprehensive multimodal feature fusion. Numerous experiments have demonstrated that the proposed SiMultiF outperforms other state-of-the-art semantic segmentation models (both multimodal and single modal) on the high-resolution ISPRS Potsdam dataset, ISPRS Vaihingen dataset, and special scene dataset (vegetation polarization dataset with extreme natural lighting contrast). Moreover, the robustness and generalizability of the network in multiscene and multimodal datasets are verified.

ZY1-02E is a medium-resolution hyperspectral operational satellite equipped with a Visible-Near Infrared Camera (VNIC), an Advanced Hyperspectral Imager (AHSI), and a Thermal Infrared Sensor (IRS). These instruments collectively provide detailed geo-morphological data that enhance environmental and resource monitoring applications. Achieving high-precision radiometric calibration is essential to fully leverage the data from ZY1-02E. This study evaluates the on-orbit radiometric performance of the ZY1-02E VNIC sensor through two calibration methods: vicarious radiometric calibration and cross-calibration. For vicarious calibration, synchronized satellite-ground observation experiments were carried out at the Dunhuang radiometric calibration site (DRCS). The reflectance-based approach was employed to validate the absolute radiometric accuracy of the VNIC. For cross-calibration, data from ZY1-02E VNIC obtained during synchronized overpasses of the DRCS were calibrated against reference sensors including Landsat-8 OLI, Landsat-9 OLI2, and Sentinel-2 MSI. Calibration coefficients were derived, and their accuracy was quantitatively evaluated. The results demonstrate that the vicarious radiometric calibration achieves an average accuracy better than 6% in the visible and near-infrared (NIR) bands, with the calibration from April showing superior performance to that of October. The average relative difference among calibration outcomes derived from various reference sensors is maintained within 2.89%, with Landsat-9 OLI2 exhibiting the closest alignment to the vicarious calibration results, showing an average relative difference of only 1.96%. Overall, the ZY1-02E VNIC sensor displays stable and excellent on-orbit radiometric performance, which is pivotal for supporting quantitative applications in environmental monitoring and resource management.

Accurate monitoring of corn phenological stages is closely related to agricultural production. By accurately monitoring the growth cycle of corn, agricultural decision-makers can more scientifically determine the timing of sowing and harvesting and maximize corn yield. Real-time phenological information on corn can provide important guidance for the implementation of crop management. Unmanned aerial vehicles (UAVs) have become an ideal platform for monitoring corn phenology because of their low cost and high spatiotemporal resolution. This study combines the advantages of the high precision and rich information of UAV images with the automation and deep feature extraction of deep learning and proposes a novel corn phenological information detection network called PhenoNet. The performance of the network is tested via UAV corn field images from five phenological stages: the seedling emergence stage, trefoil stage, jointing stage, heading stage, and mature stage. Pheno-Net adds a D-LKA deformable large kernel attention mechanism and combines the SENet channel attention mechanism with the c2f module to create c2f SENet module, which improves the detection performance of the network for different phenological stages of corn. The experimental results revealed that the accuracy, recall, and mAP50 of Pheno-Net were 89%, 92.3%, and 94.9%, respectively. The mAP50 values of the five phenological periods were 86.2%, 99%, 91.7%, 98.8%, and 98.6%, respectively. The FLOPs of the model are 3.56 G, with a size of 6.112 MB and an FPS of 334.66. Compared with current mainstream object detection networks, Pheno-Net achieves the highest accuracy, recall, detection speed, and mAP in detecting corn images of different phenological periods. This study demonstrates the enormous potential of combining unmanned aerial vehicle remote sensing technology with deep learning in corn phenological monitoring, providing a reliable phenological monitoring method for precision agriculture practices.

Interband radiometric calibration from the mid-infrared to visible bands in the ocean specular region is an effective way to calibrate on-orbit remote sensing sensors. It assumes that the referenced band has highly accurate radiance and that the interband radiometric relationship can be obtained in the ocean specular region. Most current research employs only the radiative transfer (RT) equation to derive interband radiometric relationships. However, two variables-water-leaving radiance and whitecaps-are challenging to obtain yet crucial for radiative transfer calculations. Typically, water-leaving radiance is assigned a fixed value since empirical data, whereas whitecaps are estimated via the wind speed alone. These assumptions make the uncertainties of the calibrated bands large and different from those of real satellite-measured data, reducing the reliability of the interband relationship between the reference and calibrated bands and limiting the application of the interband radiometric calibration method. To address this issue, this study proposed a novel interband radiometric calibration method called coupled deep neural networks and radiative transfer (CDR), which integrates radiative transfer and a deep neural network (DNN) to provide a reliable relationship between referenced and to be calibrated bands without accurate water-leaving radiance and whitecaps. For the four visible bands of FY-3D/MERSI-II, the relative errors were found to be 2.12%, 4.62%, 1.89%, and 4.02%, respectively. Uncertainty analysis identified the referenced band as the largest uncertainty source, followed by chlorophyll concentration, polarization effects, and aerosol loading. The CDR algorithm can be used to calibrate historical long-term satellite data without additional measurements.

Sugarcane is a widely cultivated crop in tropical regions and serves as the source of 90% of China's sugar production. In recent years, sugarcane planting in Guangxi has fluctuated due to the impacts of other cash crops and climate disasters. Real-time and accurate sugarcane mapping not only has great significance for monitoring sugarcane area and yield, but also ensures the security of sugar production. To date, there has been no research on sugarcane mapping over large areas or long time series simultaneously. In this study, we developed a mapping tool based on radar backscatter phenological information and optical indices to get accurate sugarcane mapping information. First, we analysed the distribution of sugarcane from Sentinel-1 data, extracted the typical phenological features of sugarcane from Sentinel-1 data, and then proposed the sugarcane index for sugarcane and other vegetation surfaces. Second, we combined the new radar index with the traditional optical index to obtain the 10-m resolution sugarcane distribution in Guangxi Zhuang Autonomous Region from 2018 to 2022, with an accuracy rate of 93%. Compared with the county-level statistical area, the gap was within 1%, and the R2 reached 0.99. Based on the highly accurate and stable annual sugarcane distribution map, we obtained the specific distribution locations of the new and reduced sugarcane fields each year for the first time. Our study demonstrated the feasibility of using the sugarcane index and optical vegetation index based on phenological backscattering information for sugarcane extraction, which provides annual high-precision sugarcane extraction results.

As China’s space technology rapidly progresses, the frequency of remote sensing satellite launches has significantly increased, elevating the demand for data consistency in integrated multiple sensors’ applications. Thus, the radiometric consistency of multiple sensors should be verified. The Gaofen-6 (GF-6) satellite, the first high-resolution optical remote sensing satellite dedicated to precise agricultural monitoring in China, is equipped with panchromatic and multispectral sensor (PMS) and wide field of view (WFV) and plays a pivotal role in agricultural production assessment and other related fields. In this study, evaluating the consistency of GF-6 and Sentinel-2 multispectral imager (MSI) was performed by incorporating spectral band adjustment factor and bidirectional reflectance distribution function corrections with Landsat-8 operational land imager (OLI) as the reference sensor. In addition, the radiometric consistency of top-of-atmosphere (TOA) radiance products following cross calibration was evaluated. The results were compared with those from Landsat-9 OLI2 regarding TOA radiance and surface reflectance accuracy. The fitting slopes of TOA radiance for PMS and MSI and for WFV and MSI were 0.985 and 1.038, respectively, with better consistency observed in the visible bands compared with the NIR bands. Furthermore, the comparison with Landsat-9 OLI2 TOA radiance revealed that the fitting slopes were close to 1, correlation coefficients exceeded 0.84 for both PMS and WFV, and relative errors were less than 6% . The cross-calibration approach significantly outperformed vicarious calibration in accuracy. Overall, the Landsat-8 OLI-based cross-calibration method substantially improved the consistency of TOA radiance products among PMS, WFV, and MSI sensors while maintaining stable radiometric agreement with OLI2. This study has validated the reliability and effectiveness of the OLI-based cross-calibration approach, laying a robust foundation for ensuring the consistency of multisource remote sensing data in quantitative applications.

Structured light, known for its high precision and noncontact advantages, is widely used in the fields of 3-D reconstruction and object measurement. The extraction of the light stripe center line has a significant impact on the high-precision measurement of structured light. However, traditional geometric approaches and gray centroid methods often struggle to reliably extract the stripe center, while Steger's subpixel method is hindered by intensive computational demands, making real-time applications challenging. To address these limitations while ensuring the accuracy of laser stripe center extraction, we propose a parallelization strategy for the Steger algorithm, addressing the challenges associated with high-computational load. The proposed method consists of three key components. First, a perspective projection model acts as a filter, transforming high-resolution images into a lower resolution format, thus lightening the load when identifying the region of interest (ROI). The purpose of this step is to enable laser stripe positioning. Second, we enhance the Gaussian convolutional process by implementing a separable convolutional technique, which decomposes 2-D convolution into two 1-D convolutions, thus lowering computational complexity. Finally, we adopt a dual-layer heterogeneous parallel computing mode, where laser stripe positioning and center extraction tasks are executed in parallel across CPU threads, with each thread utilizing the GPU for computation, enhancing operational efficiency by promoting in-depth collaboration between CPU and GPU. Through extensive experiments, our method demonstrates subpixel-level extraction capabilities in high-resolution laser stripe images, significantly improving execution speed while maintaining extraction accuracy. The findings indicate that the proposed approach has significant real-time application potential in the field of stripe center extraction and lays a solid foundation for improving the measurement precision of structured light.