Multi-scale Mapping Research Articles

Multiscale study of interfacial properties of carbon fiber reinforced polyphthalazine ether sulfone ketone resin matrix composites

In view of the limitations of traditional research tools on interfacial failure mechanisms in fiber/PPESK composites, this work proposes a multiscale research tool to carry out an in-depth study of the interfacial behavior between fibers and matrix. Based on microdroplet debonding tests, at the mesoscopic scale, the influence of residual thermal stress on the interface damage mode is explored through finite element (FEM) simulations. The evolution mechanism of composite material interfaces in spatial and temporal dimensions is examined based on changes in interfacial stress distribution, energy dissipation, and damage morphology during the debonding process, which can be summarized as follows: accompanied by elastic-plastic deformation and friction effects, the progressive process from localized to complete failure presents a dominant Type II damage mode at the interface. To further explore the interface failure mechanism at the molecular level, an interface model of CF/PPESK composite materials was established using molecular dynamics (MD) method. By monitoring the atom movement trend, the "fiber-matrix displacement synergistic effect" in the interfacial shear damage process was revealed, thereby establishing a multiscale mapping relationship of composite material interface. Based on this, the combination of FEM and MD was utilized to investigate the interface damage process of composite materials under different service conditions and to reasonably predict the initiation and expansion of microcracks. This study provides a pioneering perspective on interface damage research in composite materials with a "top-down" multiscale approach.

Multiscale mapping of transcriptomic signatures for cardiotoxic drugs

Drug-induced gene expression profiles can identify potential mechanisms of toxicity. We focus on obtaining signatures for cardiotoxicity of FDA-approved tyrosine kinase inhibitors (TKIs) in human induced-pluripotent-stem-cell-derived cardiomyocytes, using bulk transcriptomic profiles. We use singular value decomposition to identify drug-selective patterns across cell lines obtained from multiple healthy human subjects. Cellular pathways affected by cardiotoxic TKIs include energy metabolism, contractile, and extracellular matrix dynamics. Projecting these pathways to published single cell expression profiles indicates that TKI responses can be evoked in both cardiomyocytes and fibroblasts. Integration of transcriptomic outlier analysis with whole genomic sequencing of our six cell lines enables us to correctly reidentify a genomic variant causally linked to anthracycline-induced cardiotoxicity and predict genomic variants potentially associated with TKI-induced cardiotoxicity. We conclude that mRNA expression profiles when integrated with publicly available genomic, pathway, and single cell transcriptomic datasets, provide multiscale signatures for cardiotoxicity that could be used for drug development and patient stratification.

Multiscale rooftop greening and its socioeconomic implications in Hong Kong

Rooftop greening stands as a distinctive approach to delivering green space without occupying ground-level areas. The spatial configuration of rooftop greenery and its socioeconomic associations are critical for sustainable urban development and environmental justice. However, historical research trends have predominantly concentrated on open, ground-level environments, neglecting the vertical manifestation of urban greenery. Investigations into green roofs have typically been circumscribed to their environmental benefits, while little attention has been paid to the spatial distribution of rooftop greenery and the drivers behind it. This study aims to fill the gap by implementing a multiscale mapping approach with satellite imagery to quantify rooftop greenness in Hong Kong. An integrated analysis incorporating geographic and demographic data reveals a correlation between rooftop greenness and factors such as income levels, housing types, residential area sizes, and household compositions. This association exhibits pronounced environmental inequality, where lower-income individuals, solitary residents, and households with smaller living spaces in densely populated and aged neighborhoods are deprived of rooftop green spaces, in contrast to their counterparts in more affluent neighborhoods. Government-subsidized public housing initiatives have significantly contributed to enhancing the green living environment for low-income families. Buildings administered by government have demonstrated exceptional performance in rooftop greening efforts. This evidence provides nuanced insights into the advantages and deficiencies of green roof planning in Hong Kong and could serve as a critical reference for other densely urbanized cities working towards a sustainable and equitable urban green environment.

Complex Environment Based on Improved A* Algorithm Research on Path Planning of Inspection Robots

The proposed research aims to accomplish an improved A* algorithm for mobile robots in complex environments. In this novel algorithm, the guidance of environment information is added to the evaluation function to enhance the adaptability of the algorithm in complex environments. Additionally, to solve the problem of path smoothness, the optimal selection rules for child nodes and the bidirectional optimization strategy for path smoothing are introduced to reduce redundant nodes, which effectively makes the search space smaller and the path smoother. The simulation experiments show that, compared with the colony algorithm and Dijkstra algorithms, the proposed algorithm has significantly improved performance. Compared with the A* algorithm, the average planning time is reduced by 17.2%, the average path length is reduced by 2.05%, the average turning point is reduced by 49.4%, and the average turning Angle is reduced by 75.5%. The improved A* algorithm reduces the search space by 61.5% on average. The simulation results show that the effectiveness and adaptability of the improved A* algorithm in complex environments are verified by multi-scale mapping and multi-obstacle environment simulation experiments.

Simulation of solidification microstructure evolution of 316L stainless steel fabricated by selective laser melting using a coupled model of smooth particle hydrodynamics and cellular automata

Selective Laser Melting (SLM) is an efficient and advanced manufacturing technology that utilizes lasers to melt and solidify metal or alloy powders, layer by layer, to form the desired object shape. This study aims to investigate the evolution behavior of the solidification microstructure during the SLM process and presents a novel numerical simulation method based on the Smoothed Particle Hydrodynamics-Cellular Automaton (SPH-CA) coupling model. The SPH-CA coupling model combines the particle information, such as position and temperature, from Smoothed Particle Hydrodynamics (SPH) with the Cellular Automaton (CA) method in a one-way coupling manner, enabling simultaneous computation of multiple physical fields, including heat transfer, melting, solidification, and morphology. This model addresses challenges that are difficult to handle in grid-based computations, such as free surface variations, boundary deformations, interfacial motion, and large deformations. Additionally, it effectively simulates the morphology of the solidification microstructure. Furthermore, the coupling process performs kernel function weighted calculations at the same scale, which results in higher accuracy compared to the grid-based multiscale mapping. The simulation results demonstrate that different processing parameters influence the morphology of the molten pool and the thermal undercooling, thereby affecting the growth characteristics of the grains. A higher laser scanning speed leads to smaller molten pool size, finer grain sizes in the solidification microstructure, larger laser grain angles, and a higher content of columnar grains. These findings are consistent with experimental observations and existing conclusions, providing evidence for the feasibility and effectiveness of the proposed model in the context of SLM.

A Heuristic Approach for Resolving Spatial Conflicts of Buildings in Urban Villages

Building displacement is a common operation to resolve the spatial conflicts between map features, and it has important theoretical value and practical application significance for multi-scale mapping. The prerequisite for a successful displacement operation is that there is extra space around the conflicting buildings into which they can be displaced. Otherwise, additional generalization operators need to be combined to resolve spatial conflicts. Based on this idea, this study proposes a new heuristic spatial conflict resolution framework that mainly resolves the spatial conflicts between buildings and other features in urban villages by combining three cartographic generalization operators: selection, displacement, and aggregation. This method first reduces the density of buildings in the block through selection operation, then resolves the spatial conflicts between buildings and other features through displacement operation, and finally, the aggregation operation is performed to eliminate any remaining conflicts and newly generated conflicts. Experiments were carried out using real urban village data, and visual inspection and quantitative analysis were used to evaluate the experimental results. The evaluation results show that the proposed framework can not only resolve spatial conflicts well, but also maintain the spatial distribution and area balance of the buildings in urban villages.

Multiscale mapping and scaling analysis of the censored brittle structural framework within the crystalline bedrock of southern Finland

Fracture studies commonly lack data for the length range between 10 m to 1 km. For this reason, scaling laws are required to extrapolate fracture properties, for example in discrete fracture network models. This study focused on analysis and correlation of topology, orientation and length distribution of multiscale fracture datasets to assess their scalability. The used datasets comprise UAV-derived photogrammetric models from natural outcrops and lineaments mapped using airborne LiDAR, bathymetry and aerogeophysical data, in several contrasting scales and resolutions. This study highlights challenges in acquiring uncensored and coherent brittle structural datasets from source data characterized by a large span of resolutions between the remote sensing datasets and models of the fractured outcrop. In specific, collected data was found to be potentially biased and affected by uncertainties related to both the censoring by sedimentary cover and the scale of observation. Our results revealed differences between lineament and outcrop fracture orientations, as well as difficulties in assessing topological parameters from lineament datasets. The 1:200000 resolution was found best suited to the mapping of lineament length and resulted in a length distribution power law exponent of -1.92. For outcrop fractures that are less than 2 m long, the lognormal length distribution provided the only good fit to our data, while the longer outcrop fractures fitted relatively well with a power law exponent of -2.26.

Multi-scale mapping of Australia’s terrestrial and blue carbon stocks and their continental and bioregional drivers

The soil in terrestrial and coastal blue carbon ecosystems is an important carbon sink. National carbon inventories require accurate assessments of soil carbon in these ecosystems to aid conservation, preservation, and nature-based climate change mitigation strategies. Here we harmonise measurements from Australia’s terrestrial and blue carbon ecosystems and apply multi-scale machine learning to derive spatially explicit estimates of soil carbon stocks and the environmental drivers of variation. We find that climate and vegetation are the primary drivers of variation at the continental scale, while ecosystem type, terrain, clay content, mineralogy and nutrients drive subregional variations. We estimate that in the top 0–30 cm soil layer, terrestrial ecosystems hold 27.6 Gt (19.6–39.0 Gt), and blue carbon ecosystems 0.35 Gt (0.20–0.62 Gt). Tall open eucalypt and mangrove forests have the largest soil carbon content by area, while eucalypt woodlands and hummock grasslands have the largest total carbon stock due to the vast areas they occupy. Our findings suggest these are essential ecosystems for conservation, preservation, emissions avoidance, and climate change mitigation because of the additional co-benefits they provide.

Are open-source aerial images useful for fracture network characterisation? Insights from a multi-scale approach in the Zagros Mts.

Multi-scale fracture characterisation in reservoir analogues is crucial for determining orientation clustering, length scaling laws, abundance, and topology of the fracture network, all of which are essential for modelling flow of geofluids. In this study, we examine the application of Bing Maps imagery for fracture network analysis in the Kuh-e-Asmari anticline (Zagros Mountains, Iran), a scarcely vegetated area which is a good outcropping analogue for fractured reservoirs. Image-derived fracture characterisation is conducted at three distinct scales (1:50,000, 1:5,000 and 1:500) using the NetworkGT plugin in QGIS and was compared with structural data collected in the field (i.e. scan lines). The results reveal a scale dependency, with the advantages and disadvantages of each scale summarised as follows. The 1:50,000 is the only spatially continuous dataset. The 1:5,000 dataset, potentially spatially continuous, enables the analysis of fracture and connectivity distribution in great detail, emphasising strongly fractured/connected elongated zones, which may represent damage zones of known, previously mapped fault strands. The 1:500 scale does not ensure the spatial continuity of the dataset; however, similar to scan lines, it can perform detailed analyses of the fracture network in potentially interesting areas. We conclude that structural field surveys can be combined with multi-scale mapping on aerial images to better define the length and abundance distribution of fractures.

MMSRNet: Pathological image super-resolution by multi-task and multi-scale learning

Pathological diagnosis is the gold standard for disease assessment in clinical practice. It is conducted by inspecting the specimen at the microscopical level. Therefore, a very high-resolution pathological image that precisely describes the submicron-scale appearance is essential in the era of digital pathology, which is not easily obtained. Recently, pathological image super-resolution (SR) has shown promising prospects in bridging this gap. However, existing studies have not fully explored the peculiarity of pathological data, which contains several gradually enlarged images describing the specimen at different magnifications. In this paper, we propose a novel MMSRNet that formulates the pathological image SR in a multi-task learning way. It adds an image magnification classification branch on top of the CNN-based SR network, e.g., RCAN. Therefore, the learning objective is transformed into performing the SR while classifying the magnification as accurately as possible. The incorporated classification label guides the network to learn a more powerful feature representation. Meanwhile, the multi-task learning paradigm also encourages the joint learning of multi-scale mapping functions corresponding to multiple magnifications. It thus enables the learned model to adaptively accommodate the magnification variants, overcoming the problem that performing SR from different magnifications is treated as independent tasks in existing studies. Extensive experiments are conducted to validate the effectiveness of MMSRNet. It not only gains better performance in performing SR across magnifications and scaling factors, but also exhibits attractive plug-and-play nature when RCAN is substituted by other SR networks. The generated images are also supposed to be helpful in clinical diagnosis.

Governing Landscape in Way Khilau Micro-Catchment, Lampung Province

Indonesia has strived to combat land degradation towards viable watershed management. Watershed management at a regional scale confronts many challenges as well as socio-ecological conditions. Landscape is defined as physical boundary, a system and a holon. Landscape approach has been recognized as a multiscale mapping unit that reflects the socio-ecological dynamics within the system. Landscape services assessment offers a broader understanding of socio-ecological dynamics for local stakeholders. This study aims to provide a better understanding of building spatial decision-making through the landscape approach at Way Khilau micro-catchment scale. The study area is about 1.162 Ha and is located in the upper basin of Way Bulog, Lampung Province, Indonesia. Driver-Pressure-Impact-Response (DPSIR) framework was used to map the landscape services within the watershed qualitatively. The result showed major physical drivers such as relief configuration took control on the hydrological services. Way Khilau micro-catchment arranged by parallel-narrow ridge, undulating to very steep slope, and valley. Both ridge and slope played a role for the hydrological regulation services and the lowest part provided water resources for the socio-economic activity. Further research of environmental changes needs to be assessed quantitatively to determine the impact of stakeholder’s responses on socio-ecological issues.

Schizophrenia Polygenic Risk During Typical Development Reflects Multiscale Cortical Organization

BackgroundSchizophrenia is widely recognized as a neurodevelopmental disorder. Abnormal cortical development in otherwise typically developing children and adolescents may be revealed using polygenic risk scores for schizophrenia (PRS-SCZ). MethodsWe assessed PRS-SCZ and cortical morphometry in typically developing children and adolescents (3–21 years, 46.8% female) using whole-genome genotyping and T1-weighted magnetic resonance imaging (n = 390) from the PING (Pediatric Imaging, Neurocognition, and Genetics) cohort. We contextualized the findings using 1) age-matched transcriptomics, 2) histologically defined cytoarchitectural types and functionally defined networks, and 3) case-control differences of schizophrenia and other major psychiatric disorders derived from meta-analytic data of 6 ENIGMA (Enhancing Neuro Imaging Genetics through Meta Analysis) working groups, including a total of 12,876 patients and 15,670 control participants. ResultsHigher PRS-SCZ was associated with greater cortical thickness, which was most prominent in areas with heightened gene expression of dendrites and synapses. PRS-SCZ–related increases in vertexwise cortical thickness were mainly distributed in association cortical areas, particularly the ventral attention network, while relatively sparing koniocortical type cortex (i.e., primary sensory areas). The large-scale pattern of cortical thickness increases related to PRS-SCZ mirrored the pattern of cortical thinning in schizophrenia and mood-related psychiatric disorders derived from the ENIGMA consortium. Age group models illustrate a possible trajectory from PRS-SCZ–associated cortical thickness increases in early childhood toward thinning in late adolescence, with the latter resembling the adult brain phenotype of schizophrenia. ConclusionsCollectively, combining imaging genetics with multiscale mapping, our work provides novel insight into how genetic risk for schizophrenia affects the cortex early in life.

Multi-scale mapping algorithm for INI cancellation and a novel mixed-numerologies OFDM transceiver

OFDM with mixed-numerologies enhances the system flexibility effectively to meet the demands of diversified application scenarios. However, the coexistence of waveforms with different numerologies leads to serious inter-numerology interference (INI), and the corresponding relationship between the number of guard subcarriers and the power of INI needs to be considered for scheduling subcarriers. In this paper, we propose a multi-scale mapping (MSM) and INI cancellation (MSM-INIC) algorithm as well as the corresponding de-MSM algorithm for mixed-numerologies OFDM system. Based on the proposed algorithms, we provide a novel transceiver in the scenario of multi-path fading channel, in which subcarrier scheduling does not need to consider whether the guard band is allocated. In the proposed transmitter, an additional MSM-INIC module is employed to pre-compensate signal distortion for downlink, and in the receiver, a de-MSM module is applied to de-map the received signals for recovering the original numerologies. Furthermore, we reveal the inherent property of the mapped signals, and propose a low computational complexity de-MSM algorithm accordingly. Simulation results verify the superiority of the proposed transceiver in BER performance as well as spectrum efficiency even without any guard band.

Decoupling Slope and Aspect Vectors to Generalize Relief Shading

ABSTRACT Relief shading is designed to vary the brightness of terrain elements on a two-dimensional map to create a three-dimensional effect. One concern is how this layer can be generalized for use in multi-scale mapping. We propose a methodology that calculates relief shading from slope and aspect vectors, as these layers allow map users to recognize characteristics of the terrain and show consistent trends in spatial autocorrelation with generalization. We adjust the orientation of surface vectors with a mean filter to preserve the structural terrain elements while eliminating landforms of finer detail. To demonstrate its use, we show two examples of generalizing detailed relief shading and compare results to relief shading of the next coarser scale of DEM data available. The generalized maps remove or smooth out minor landforms while preserving more prominent landforms and eliminate issues of data gaps or interpolated data in lower resolution datasets.

Multiscale structural mapping of Alzheimer’s disease neurodegeneration

The recently described biological framework of Alzheimer’s disease (AD) emphasizes three types of pathology to characterize this disorder, referred to as the ‘amyloid/tau/neurodegeneration’ (A-T-N) status. The ‘neurodegenerative’ component is typically defined by atrophy measures derived from structural magnetic resonance imaging (MRI) such as hippocampal volume. Neurodegeneration measures from imaging are associated with disease symptoms and prognosis. Thus, sensitive image-based quantification of neurodegeneration in AD has an important role in a range of clinical and research operations. Although hippocampal volume is a sensitive metric of neurodegeneration, this measure is impacted by several clinical conditions other than AD and therefore lacks specificity. In contrast, selective regional cortical atrophy, known as the ‘cortical signature of AD’ provides greater specificity to AD pathology. Although atrophy is apparent even in the preclinical stages of the disease, it is possible that increased sensitivity to degeneration could be achieved by including tissue microstructural properties in the neurodegeneration measure. However, to facilitate clinical feasibility, such information should be obtainable from a single, short, noninvasive imaging protocol. We propose a multiscale MRI procedure that advances prior work through the quantification of features at both macrostructural (morphometry) and microstructural (tissue properties obtained from multiple layers of cortex and subcortical white matter) scales from a single structural brain image (referred to as ‘multi-scale structural mapping’; MSSM). Vertex-wise partial least squares (PLS) regression was used to compress these multi-scale structural features. When contrasting patients with AD to cognitively intact matched older adults, the MSSM procedure showed substantially broader regional group differences including areas that were not statistically significant when using cortical thickness alone. Further, with multiple machine learning algorithms and ensemble procedures, we found that MSSM provides accurate detection of individuals with AD dementia (AUROC = 0.962, AUPRC = 0.976) and individuals with mild cognitive impairment (MCI) that subsequently progressed to AD dementia (AUROC = 0.908, AUPRC = 0.910). The findings demonstrate the critical advancement of neurodegeneration quantification provided through multiscale mapping. Future work will determine the sensitivity of this technique for accurately detecting individuals with earlier impairment and biomarker positivity in the absence of impairment.

An End-to-End Traffic Visibility Regression Algorithm

Traffic visibility detection plays a vital role in intelligent transportation, autonomous driving, safe driving, etc. Convolutional neural networks (CNNs) based regression and classification algorithms have been shown competitive performance in many applications, but little attention has been paid to traffic visibility identification. In this paper, we propose a trainable end-to-end system called traffic visibility regression network (TVRNet). TVRNet takes a road image as input and outputs its visibility value. TVRNet adopts CNNs based deep architecture, uses appropriate filters to extract fog density-related features, and exploits the parallel convolution for multi-scale mapping. Later, a new type of non-linear activation function called Modified_sigmoid function is used. We synthesize labeled visibility datasets comprised of multi-scene and single-scene based on the actual road sense to train the visibility regression network. Extensive experiments and comparisons with other popular algorithms are performed to verify our method in road visibility estimation.

Random Topology and Random Multiscale Mapping: An Automated Design of Multiscale and Lightweight Neural Network for Remote-Sensing Image Recognition

With the proposal of neural architecture search (NAS), automated network architecture design gradually becomes a new way in deep learning research. Due to its high capability regarding automated design, some pioneers have made an attempt to apply NAS in remote sensing and made some achievements, like 1-D/3-D Auto-convolutional neural network (CNN) and polarimetric synthetic aperture radar (PolSAR)-tailored Differentiable Architecture Search (PDAS). However, there are still some areas to be improved for existing NAS in remote-sensing field. In this article, we propose a random topology and random multiscale mapping (RTRMM) method to generate a multiscale and lightweight architecture for remote-sensing image recognition. First, a random topology generator generates the topology through random graph. Second, during the experiment, we find remote-sensing image features extracted by a multiscale network are more appropriate, compared with features extracted by a single-scale model. Nevertheless, the complexity inevitably increases with the introduction of a multiscale concept. Consequently, we design a variable search space consisting of decomposition convolution units under the guidance of mathematical analysis. The mapping of each neuron is then determined by a random multiscale mapping sampler. After that, we assemble the topology and mappings into blocks and construct three RTRMM models. Experiments on four scene classification datasets confirm the feature extraction capability and lightweight performance of RTRMM models. Moreover, we also observe that our approach achieves a better tradeoff between floating-point operations (FLOPs) and accuracy than some current well-behaved methods. Furthermore, the results on Vaihingen dataset verify the high feature-transfer capability.

Tracing episodic accretion with NOEMA: FU Orionis-type stars and their millimeter environment

The earliest phases of star formation are characterised by intense mass accretion from the circumstellar disk to the central star. One group of lowmass young stellar objects, FU Orionis-type stars (FUors) exhibit accretion rate peaks accompanied by bright eruptions in the optical and infrared regime. The occurance of these outbursts might solve the luminosity problem of protostars, play a key role in accumulating the final stellar mass, and have a significant effiect on the parameters of the envelope and the disk. We are performing a systematic investigation of FUors with millimeter interferometry using NOEMA and ALMA to study the outburst events and examine whether FUors represent normal young stars in exceptional times or are unusual objects. The targeted FUors show very diverse circumstellar morphologies with envelope parameters similar to those of both Class I and Class II systems, but their disks are more massive and more compact than T Tauri disks. To shed light onto the process of disk-formation, accretion, and to what role FUors play in low-mass starformation, we require the identification and light curve monitoring of as many of these stars as possible, together with the multi-wavelength and multi-scale mapping of their circumstellar environment.

Computer-Aided Diagnosis for Pneumoconiosis Staging Based on Multi-scale Feature Mapping

In this research, we explored a method of multi-scale feature mapping to pre-screen radiographs quickly and accurately in the aided diagnosis of pneumoconiosis staging. We utilized an open dataset and a self-collected dataset as research datasets. We proposed a multi-scale feature mapping model based on deep learning feature extraction technology for detecting pulmonary fibrosis and a discrimination method for pneumoconiosis staging. The diagnostic accuracy was evaluated using under the curve (AUC) of the receiver operating characteristic (ROC) curve. The AUC value of our model was 0.84, which showed the best performance compared with previous work on datasets. The diagnosis results indicated that our method was highly consistent with that of clinical experts on real patient. Furthermore, the AUC value obtained through categories I–IV on the testing dataset demonstrated that categories I (AUC = 0.86) and IV (AUC = 0.82) obtained the best performance and achieved the level of clinician categorization. Our research could be applied to the pre-screening and diagnosis of pneumoconiosis in the clinic.

TWC-Net: A SAR Ship Detection Using Two-Way Convolution and Multiscale Feature Mapping

Synthetic aperture radar (SAR) is an active earth observation system with a certain surface penetration capability and can be employed to observations all-day and all-weather. Ship detection using SAR is of great significance to maritime safety and port management. With the wide application of in-depth learning in ordinary images and good results, an increasing number of detection algorithms began entering the field of remote sensing images. SAR image has the characteristics of small targets, high noise, and sparse targets. Two-stage detection methods, such as faster regions with convolution neural network (Faster RCNN), have good results when applied to ship target detection based on the SAR graph, but their efficiency is low and their structure requires many computing resources, so they are not suitable for real-time detection. One-stage target detection methods, such as single shot multibox detector (SSD), make up for the shortage of the two-stage algorithm in speed but lack effective use of information from different layers, so it is not as good as the two-stage algorithm in small target detection. We propose the two-way convolution network (TWC-Net) based on a two-way convolution structure and use multiscale feature mapping to process SAR images. The two-way convolution module can effectively extract the feature from SAR images, and the multiscale mapping module can effectively process shallow and deep feature information. TWC-Net can avoid the loss of small target information during the feature extraction, while guaranteeing good perception of a large target by the deep feature map. We tested the performance of our proposed method using a common SAR ship dataset SSDD. The experimental results show that our proposed method has a higher recall rate and precision, and the F-Measure is 93.32%. It has smaller parameters and memory consumption than other methods and is superior to other methods.