Angle Mapper Research Articles

Operational use of multispectral images for macro-litter mapping and categorization by Unmanned Aerial Vehicle

The use of Unmanned Aerial Systems (UAS, aka drones) has shown to be feasible to perform marine litter surveys. We operationally tested the use of multispectral images (5 bands) to classify litter type and material on a beach-dune system. For litter categorization by their multispectral characteristics, the Spectral Angle Mapping (SAM) technique was adopted.The SAM-based categorization of litter agreed with the visual classification, thus multispectral images can be used to fasten and/or making more robust the manual RGB image screening. Fully automated detection returned an F-score of 0.64, and a reasonable categorization of litter. Overall, the image-based litter density maps were in line with the manual detection. Assessments were promising given the complexity of the study area, where different dunes plants and partially-buried items challenged the UAS-based litter detection. The method can be easily implemented for both floating and beached litter, to advance litter survey in the environment.

Application of Remote Sensing Tools to Assess the Land Use and Land Cover Change in Coatzacoalcos, Veracruz, Mexico

Land use and land cover (LULC) change has become an important research topic for global environmental change and sustainable development. As an important part of worldwide land conservation, sustainable development and management of water resources, developing countries must ensure the use of innovative technology and tools that support their various decision making systems. This study provides the most recent LULC change analysis for the last six years (2015–2021) of Coatzacoalcos, Veracruz, Mexico, one of the most important petrochemical cities in the world and host of the ongoing Interoceanic Corridor project. The analysis was carried out using Landsat 8 Operational Land Imager (OLI) satellite images, ancillary data and ground-based surveys and the Normalized Difference Vegetation Index (NDVI) to identify and to ameliorate the discrimination between four main macro-classes and fourteen classes. The LULC classification was performed using the maximum likelihood classifier (MLC) to produce maps for each year, as it was found to be the best approach when compared to minimum distance (MDM) and spectral angle mapping (SAM) methods. The macro-classes were water, built-up, vegetation and bare soil, whereas the classes were an improved classification within those. Our study achieved both user accuracy (UA) and producer accuracy (PA) above 90% for the proposed macro-classes and classes. The average Kappa coefficient for macro-classes was 0.93, while for classes it was 0.96, both comparable to previous studies. The results from the LULC analysis show that residential, industry and commercial areas slowed down their growth throughout the study period. These changes were associated with socio-economical drivers such as insecurity and lack of economic investments. Groves and trees presented steady behaviors, with small increments during the five-year period. Swamps, on the other hand, significantly degraded, being about 2% of the study area in 2015 and 0.93% in 2021. Dunes and medium and high vegetation densities (∼80%) transitioned mostly to low vegetation densities. This behavior is associated with rainfall below the annual reference and increments of surface runoff due to the loss of vegetation cover. Lastly, the present study seeks to highlight the importance of remote sensing for a better understanding of the dynamics between human–nature interactions and to provide information to assist planners and decision-makers for more sustainable land development.

Characterizing the macroscopic response and local deformation mechanisms of a unidirectional non-crimp fabric

The macroscopic response and local deformation mechanisms of a warp stitched heavy-tow unidirectional non-crimp fabric (UD-NCF) was characterized. Quasi-static uniaxial tests captured the fabric in-plane longitudinal, transverse, and shear deformation response where digital image correlation was employed. The fabric bending response was captured through modified cantilever tests. Off-axis extension tests conducted along 30°, 45° and 60° bias directions revealed that extension and contraction of the polyester stitching web and resulting transverse compression of the carbon fiber tows and inter-tow sliding were the dominant local deformation mechanisms. At a bias angle of 30°, the extent of inter-tow sliding increased when compared to a bias angle of 45° leading to earlier onset of fabric shear locking, while at a bias angle of 60° shear locking also occurred sooner. Shear angle maps and strain profiles at different bias angles further revealed the notable inter-dependency between the fabric shear, longitudinal and transverse deformations, which was governed by the interactions between the stitching and carbon fiber tows. The distinct out-of-plane bending behavior of the fabric reinforcement along longitudinal and transverse directions was also captured and quantified. The study provided a comprehensive characterization of the UD-NCF, which is critical for the development of constitutive models.

Fully automated quantification of in vivo viscoelasticity of prostate zones using magnetic resonance elastography with Dense U-net segmentation

Magnetic resonance elastography (MRE) for measuring viscoelasticity heavily depends on proper tissue segmentation, especially in heterogeneous organs such as the prostate. Using trained network-based image segmentation, we investigated if MRE data suffice to extract anatomical and viscoelastic information for automatic tabulation of zonal mechanical properties of the prostate. Overall, 40 patients with benign prostatic hyperplasia (BPH) or prostate cancer (PCa) were examined with three magnetic resonance imaging (MRI) sequences: T2-weighted MRI (T2w), diffusion-weighted imaging (DWI), and MRE-based tomoelastography, yielding six independent sets of imaging data per patient (T2w, DWI, apparent diffusion coefficient, MRE magnitude, shear wave speed, and loss angle maps). Combinations of these data were used to train Dense U-nets with manually segmented masks of the entire prostate gland (PG), central zone (CZ), and peripheral zone (PZ) in 30 patients and to validate them in 10 patients. Dice score (DS), sensitivity, specificity, and Hausdorff distance were determined. We found that segmentation based on MRE magnitude maps alone (DS, PG: 0.93 ± 0.04, CZ: 0.95 ± 0.03, PZ: 0.77 ± 0.05) was more accurate than magnitude maps combined with T2w and DWI_b (DS, PG: 0.91 ± 0.04, CZ: 0.91 ± 0.06, PZ: 0.63 ± 0.16) or T2w alone (DS, PG: 0.92 ± 0.03, CZ: 0.91 ± 0.04, PZ: 0.65 ± 0.08). Automatically tabulated MRE values were not different from ground-truth values (P>0.05). In conclusion, MRE combined with Dense U-net segmentation allows tabulation of quantitative imaging markers without manual analysis and independent of other MRI sequences and can thus contribute to PCa detection and classification.

Study on the impedance characteristics of a high-capacity pulse tube cryocooler

Stirling-type pulse tube cryocoolers are used in many applications such as military facilities, aerospace devices and superconductive systems. A major problem in improving its energy conversion efficiency is how to achieve a better impedance matching. In this study, a pulse tube cryocooler is investigated based on the impedance analyses. A group of impedance maps is proposed to analyze the energy conversion performance of the compressor. The velocity map, the efficiency map, the acoustic power map and the phase angle map are provided under the total equivalent impedance. These maps offer a visualized guidance of the maximum performance a compressor could reach and in which direction to optimize. Experiments on a high capacity pulse tube cryocooler are conducted to verify the theoretical analyses. The results show that it is of great help in using the proposed maps to analyze and improve the energy conversion performance of the cryocooler. A cooling capacity of 300W@80K is finally achieved. The study clearly reveals the influence of the impedance on the energy conversion efficiency and it is quite useful in improving the performance of the pulse tube cryocooler.

MP3RAGE: Simultaneous mapping of T1 and in human brain at 7T.

To map T1 and the local flip angle ( ) in human brain using a single MP3RAGE sequence with 3 rapid acquisitions of gradient echoes (RAGEs). A third RAGE with a relatively high flip angle was appended to an MP2RAGE sequence. Through curve fitting and a rational approximation for small flip angles and short TR, closed form solutions for T1 and were derived. The influence of different k-space encoding schemes on precision and whether edge enhancement artifacts could be reduced with a saturation pulse applied prior to the third RAGE were explored. Validation of T1 estimates was performed using single-slice inversion recovery (IR) and a subsequent region-of-interest-based comparison, whereas validation of was performed using a whole brain pixelwise comparison to a DREAM flip angle mapping protocol. Lastly, MP3RAGE was compared to T1 -mapping by MP2RAGE with separate correction. Whole brain maps of T1 and at 1 mm isotropic resolution were obtained with MP3RAGE in 06:37 min. A linear-reverse centric-reverse centric phase-encoding order of the 3 RAGEs improved precision, and artifacts were successfully reduced with the saturation pulse. Estimations of T1 and deviated +2.5 ± 3.1% and -1.7 ± 8.6% from their respective references. T1 and can be mapped simultaneously using MP3RAGE. The approach can be thought of as combining MP2RAGE with a dual flip angle T1 -mapping protocol. Both maps can be solved for analytically and will be inherently co-registered at the high resolution associated with MPRAGE.

Detection of alteration zones using the Dirichlet process Stick-Breaking model-based clustering algorithm to hyperion data: the case study of Kuh-Panj porphyry copper deposits, Southern Iran

Detection of hydrothermal alteration zones (HAZs) associated with porphyry copper systems using remote sensing imagery is a crucial stage for discovering high potential zone of ore mineralization. Statistical model-based clustering methods have great potential for automatic and accurate detection of hydrothermal alteration minerals using hyperspectral remote sensing imagery. In this research, the Dirichlet Process based on Stick-Breaking (DPSB) model-based clustering algorithm was implemented to hyperion remote sensing imagery to discriminate HAZs associated with the Kuh-Panj porphyry copper deposit, south, Iran. The DPSB clustering algorithm was implemented and subsequently compared with the k-means algorithm, CLARA clustering, hierarchical clustering, Gaussian finite mixture model (GFMM), Gaussian model for high-dimensional (GMHD) and spectral clustering as well as spectral angle mapping (SAM). Results derived from the DPSB model-based clustering algorithm show 88.6% accuracy in distinguishing propylitic, argillic, advanced argillic, propylitic-argillic and phyllic alteration zones. The DPSB algorithm can be broadly implemented to hyperspectral remote sensing imagery for detecting alteration zones associated with porphyry systems.

Improving LSMA for impervious surface estimation in an urban area

ABSTRACT Linear spectral mixture analysis (LSMA) and regression analysis are the two most conventionally used methods to estimate impervious surfaces at the subpixel scale in an urban area. However, LSMA lacks the sensitivity to pixel brightness, which leads to inter variability of endmembers and affects the ability to distinguish features with a similar spectral signature. This research aims to develop LSMA aided by a regression analysis model to estimate impervious surfaces with higher accuracy. A spectral angle mapping (SAM) based regression analysis model is introduced to reduce errors. Based on high-resolution images and field survey data, the SAM-based regression analysis can estimate non-impervious surface and high-impervious surface densities with high accuracy, while less accurate in impervious surfaces with low/medium density. In contrast, LSMA is able to estimate low/medium-density impervious surfaces with higher accuracy. We propose an improved approach by integrating the two methods, regression analysis aided LSMA, for impervious surface estimation. The proposed method increases the overall accuracy of the impervious surface estimation to 85.24%, which is significantly greater than that of the conventional methods.

K-means find density peaks in molecular conformation clustering

Performing cluster analysis on molecular conformation is an important way to find the representative conformation in the molecular dynamics trajectories. Usually, it is a critical step for interpreting complex conformational changes or interaction mechanisms. As one of the density-based clustering algorithms, find density peaks (FDP) is an accurate and reasonable candidate for the molecular conformation clustering. However, facing the rapidly increasing simulation length due to the increase in computing power, the low computing efficiency of FDP limits its application potential. Here we propose a marginal extension to FDP named K-means find density peaks (KFDP) to solve the mass source consuming problem. In KFDP, the points are initially clustered by a high efficiency clustering algorithm, such as K-means. Cluster centers are defined as typical points with a weight which represents the cluster size. Then, the weighted typical points are clustered again by FDP, and then are refined as core, boundary, and redefined halo points. In this way, KFDP has comparable accuracy as FDP but its computational complexity is reduced from O(n2) to O(n). We apply and test our KFDP method to the trajectory data of multiple small proteins in terms of torsion angle, secondary structure or contact map. The comparing results with K-means and density-based spatial clustering of applications with noise show the validation of the proposed KFDP.

The relationship between myocardial microstructure and strain in chronic infarction using cardiovascular magnetic resonance diffusion tensor imaging and feature tracking

BackgroundCardiac diffusion tensor imaging (cDTI) using cardiovascular magnetic resonance (CMR) is a novel technique for the non-invasive assessment of myocardial microstructure. Previous studies have shown myocardial infarction to result in loss of sheetlet angularity, derived by reduced secondary eigenvector (E2A) and reduction in subendocardial cardiomyocytes, evidenced by loss of myocytes with right-handed orientation (RHM) on helix angle (HA) maps. Myocardial strain assessed using feature tracking-CMR (FT-CMR) is a sensitive marker of sub-clinical myocardial dysfunction. We sought to explore the relationship between these two techniques (strain and cDTI) in patients at 3 months following ST-elevation MI (STEMI).Methods32 patients (F = 28, 60 ± 10 years) underwent 3T CMR three months after STEMI (mean interval 105 ± 17 days) with second order motion compensated (M2), free-breathing spin echo cDTI, cine gradient echo and late gadolinium enhancement (LGE) imaging. HA maps divided into left-handed HA (LHM, − 90 < HA < − 30), circumferential HA (CM, − 30° < HA < 30°), and right-handed HA (RHM, 30° < HA < 90°) were reported as relative proportions. Global and segmental analysis was undertaken.ResultsMean left ventricular ejection fraction (LVEF) was 44 ± 10% with a mean infarct size of 18 ± 12 g and a mean infarct segment LGE enhancement of 66 ± 21%. Mean global radial strain was 19 ± 6, mean global circumferential strain was − 13 ± − 3 and mean global longitudinal strain was − 10 ± − 3. Global and segmental radial strain correlated significantly with E2A in infarcted segments (p = 0.002, p = 0.011). Both global and segmental longitudinal strain correlated with RHM of infarcted segments on HA maps (p < 0.001, p = 0.003). Mean Diffusivity (MD) correlated significantly with the global infarct size (p < 0.008). When patients were categorised according to LVEF (reduced, mid-range and preserved), all cDTI parameters differed significantly between the three groups.ConclusionChange in sheetlet orientation assessed using E2A from cDTI correlates with impaired radial strain. Segments with fewer subendocardial cardiomyocytes, evidenced by a lower proportion of myocytes with right-handed orientation on HA maps, show impaired longitudinal strain. Infarct segment enhancement correlates significantly with E2A and RHM. Our data has demonstrated a link between myocardial microstructure and contractility following myocardial infarction, suggesting a potential role for CMR cDTI to clinically relevant functional impact.

Cover Image

COVER ILLUSTRATION Seeing the invisible – Intraoperative Optical Imaging (IOI) is a technique that gives us unique insights on the human brain and it's function. The method uses subtle changes in the optical properties of the exposed cortical brain tissue to detect and visualize hemodynamic changes which are connected to specific brain functions. Since IOI needs beside an undisturbed view on the cortical surface only a camera system and light within the visible range it is well suited for application during neurosurgical interventions. In our paper “Characterization of Cortical Hemodynamic Changes Following Sensory, Visual, and Speech Activation by Intraoperative Optical Imaging Utilizing Phase-Based Evaluation Methods” we used IOI to characterize the hemodynamic changes on the brain surface following different types of stimulation in patients that underwent brain tumor surgery. The graphics illustrates the several aspects of the methodology and the results of our paper. The three images in the middle are showing phase angle maps, overlaid to the white light image of the exposed brain surfaces. The different maps are representing the response towards visual (left), sensory (right), and speech activation (middle) as shown in the result section of the paper and supported by the text above the whole image composition. The grid in the front is a polar histogram, we used for quantitative analysis and visualization of the results. Overlaid to the grid is the colorbar that we used for the phase angle maps, shaped as vortex. The colors correspond with different phase angles of the polar histogram. To underline the fact that IOI can be used to visualize functional brain areas during neurosurgical interventions, we added in the back an abstract painting of a neurosurgeon, looking through the sterile draped surgical microscope. The light source references to the fact, that IOI is an optical technology and therefore based on light.

How 2.5D Maps Design Improve the Wayfinding Performance and Spatial Ability of Map Users

Useful information can be provided by 2.5D maps that can take advantage of the additional dimension. However, aside from stereoscopic landmarks, optimal methods for presenting other essential information is unclear. Two experiments were conducted to explore how the presentation of 2.5D maps can effectively increase wayfinding performance. First, analysis was performed to understand the effects of 2.5D maps on wayfinding behavior and map reading. Then, a 2.5D map design was proposed and verified to optimize the 2.5D map presentation of urban environments. The results showed that compared with users of low view angle maps, those using high view angle maps orientated more easily with elements of the map during wayfinding tasks. High view angle maps allowed superior performance, and including transparency and lines improved wayfinding performance. The participants using maps that were opaque and with lines exhibited the most confusion and hesitation. The participants who used maps that were transparent and had lines exhibited the least confusion and hesitation. Highlighting buildings at intersections can help map users use the intersections as references and increase their intuitive spatial ability.

Orthogonal decomposition of composition operators on the H2 space of Dirichlet series

Let H2 denote the Hilbert space of Dirichlet series with square-summable coefficients. We study composition operators Cφ on H2 which are generated by symbols of the form φ(s)=c0s+∑n≥1cnn−s, in the case that c0≥1. If only a subset P of prime numbers features in the Dirichlet series of φ, then the operator Cφ admits an associated orthogonal decomposition. Under sparseness assumptions on P we use this to asymptotically estimate the approximation numbers of Cφ. Furthermore, in the case that φ is supported on a single prime number, we affirmatively settle the problem of describing the compactness of Cφ in terms of the ordinary Nevanlinna counting function. We give detailed applications of our results to affine symbols and to angle maps.

A Rapid Model (COV_PSDI) for Winter Wheat Mapping in Fallow Rotation Area Using MODIS NDVI Time-Series Satellite Observations: The Case of the Heilonggang Region

Rapid and accurate monitoring of spatial distribution patterns of winter wheat over a long period is of great significance for crop yield prediction and farmland water consumption estimation. However, weather conditions and relatively long revisit cycles often result in an insufficient number of continuous medium-high resolution images over large areas for many years. In addition, the cropland pattern changes frequently in the fallow rotation area. A novel rapid mapping model for winter wheat based on the normalized difference vegetation index (NDVI) time-series coefficient of variation (NDVI_COVfp) and peak-slope difference index (PSDI) is proposed in this study. NDVI_COVfp uses the time-series index volatility to distinguish cultivated land from background land-cover types. PSDI combines the key growth stages of winter wheat phenology and special bimodal characteristics, substantially reducing the impact of abandoned land and other crops. Taking the Heilonggang as an example, this study carried out a rapid mapping of winter wheat for four consecutive years (2014–2017), and compared the proposed COV_PSDI with two state-of-the-art methods and traditional methods (the Spectral Angle Mapping (SAM) and the Iterative Self-Organizing Data Analysis Technique Algorithm (ISODATA)). The verification results revealed that the COV_PSDI model improved the overall accuracy (94.10%) by 4% compared with the two state-of-art methods (90.80%, 89.00%) and two traditional methods (90.70%, 87.70%). User accuracy was the highest, which was 93.74%. Compared with the other four methods, the percentage error (PE) of COV_PSDI for four years was the lowest in the same year, with the minimum variation range of PE being 1.6–3.6%. The other methods resulted in serious overestimation. This demonstrated the effectiveness and stability of the method proposed in the rapid and accurate extraction of winter wheat in a large area of fallow crop rotation region. Our study provides insight for remote sensing monitoring of spatiotemporal patterns of winter wheat and evaluation of “fallow rotation” policy implementation.

Application of Maximum Likelihood and Spectral Angle Mapping Classification Techniques to Evaluate Forest Fire Severity from UAV Multi-spectral Images in South Korea

Application of Maximum Likelihood and Spectral Angle Mapping Classification Techniques to Evaluate Forest Fire Severity from UAV Multi-spectral Images in South Korea

Polarization Tomography with Stokes Parameters

We present a simple but powerful technique for the analysis of polarized emission from radio galaxies and other objects. It is based on the fact that images of Stokes parameters often contain considerably more information than is available in polarized intensity and angle maps. In general, however, the orientation of the Stokes parameters will not be matched to the position angles of structures in the source. Polarization tomography, the technique presented in this paper, consists of making a series of single linear Stokes parameter images, S(ρ), where each image is rotated by an angle ρ from the initial orientation of Q and U. Examination of these images, in a series of still frames or a movie, reveals often hidden patterns of polarization angles, as well as structures that were obscured by the presence of overlapping polarized emission. We provide both cartoon examples and a quick look at the complex polarized structure in Cygnus A.

Learned Hyperspectral Compression Using a Student’s T Hyperprior

Hyperspectral compression is one of the most common techniques in hyperspectral image processing. Most recent learned image compression methods have exhibited excellent rate-distortion performance for natural images, but they have not been fully explored for hyperspectral compression tasks. In this paper, we propose a trainable network architecture for hyperspectral compression tasks, which not only considers the anisotropic characteristic of hyperspectral images but also embeds an accurate entropy model using the non-Gaussian prior knowledge of hyperspectral images and nonlinear transform. Specifically, we first design a spatial-spectral block, involving a spatial net and a spectral net as the base components of the core autoencoder, which is more consistent with the anisotropic hyperspectral cubes than the existing compression methods based on deep learning. Then, we design a Student’s T hyperprior that merges the statistics of the latents and the side information concepts into a unified neural network to provide an accurate entropy model used for entropy coding. This not only remarkably enhances the flexibility of the entropy model by adjusting various values of the degree of freedom, but also leads to a superior rate-distortion performance. The results illustrate that the proposed compression scheme supersedes the Gaussian hyperprior universally for virtually all learned natural image codecs and the optimal linear transform coding methods for hyperspectral compression. Specifically, the proposed method provides a 1.51% to 59.95% average increase in peak signal-to-noise ratio, a 0.17% to 18.17% average increase in the structural similarity index metric and a 6.15% to 64.60% average reduction in spectral angle mapping over three public hyperspectral datasets compared to the Gaussian hyperprior and the optimal linear transform coding methods.

Radiofrequency Bias Correction of Magnetization Prepared Rapid Gradient Echo MRI at 7.0 Tesla Using an External Reference in a Sequential Protocol.

At field strengths of 7 T and above, T1-weighted imaging of human brain suffers increasingly from radiofrequency (RF) B1 inhomogeneities. The well-known MP2RAGE (magnetization prepared two rapid acquisition gradient echoes) sequence provides a solution but may not be readily available for all MR systems. Here, we describe the implementation and evaluation of a sequential protocol to obtain normalized magnetization prepared rapid gradient echo (MPRAGE) images at 0.7, 0.8, or 0.9-mm isotropic spatial resolution. Optimization focused on the reference gradient-recalled echo (GRE) that was used for normalization of the MPRAGE. A good compromise between white-gray matter contrast and the signal-to-noise ratio (SNR) was reached at a flip angle of 3° and total scan time was reduced by increasing the reference voxel size by a factor of 8 relative to the MPRAGE resolution. The average intra-subject coefficient-of-variation (CV) in segmented white matter (WM) was 7.9 ± 3.3% after normalization, compared to 20 ± 8.4% before. The corresponding inter-subject average CV in WM was 7.6 ± 7.6% and 13 ± 7.8%. Maps of T1 derived from forward signal modelling showed no obvious bias after correction by a separately acquired flip angle map. To conclude, a non-interleaved acquisition for normalization of MPRAGE offers a simple alternative to MP2RAGE to obtain semi-quantitative purely T1-weighted images. These images can be converted to T1 maps, analogously to the established MP2RAGE approach. Scan time can be reduced by increasing the reference voxel size which has only a miniscule effect on image quality.

Artificial Intelligence for Pigment Classification Task in the Short-Wave Infrared Range.

Hyperspectral reflectance imaging in the short-wave infrared range (SWIR, “extended NIR”, ca. 1000 to 2500 nm) has proven to provide enhanced characterization of paint materials. However, the interpretation of the results remains challenging due to the intrinsic complexity of the SWIR spectra, presenting both broad and narrow absorption features with possible overlaps. To cope with the high dimensionality and spectral complexity of such datasets acquired in the SWIR domain, one data treatment approach is tested, inspired by innovative development in the cultural heritage field: the use of a pigment spectral database (extracted from model and historical samples) combined with a deep neural network (DNN). This approach allows for multi-label pigment classification within each pixel of the data cube. Conventional Spectral Angle Mapping and DNN results obtained on both pigment reference samples and a Buddhist painting (thangka) are discussed.

R-CenterNet+: Anchor-Free Detector for Ship Detection in SAR Images.

In recent years, the rapid development of Deep Learning (DL) has provided a new method for ship detection in Synthetic Aperture Radar (SAR) images. However, there are still four challenges in this task. (1) The ship targets in SAR images are very sparse. A large number of unnecessary anchor boxes may be generated on the feature map when using traditional anchor-based detection models, which could greatly increase the amount of computation and make it difficult to achieve real-time rapid detection. (2) The size of the ship targets in SAR images is relatively small. Most of the detection methods have poor performance on small ships in large scenes. (3) The terrestrial background in SAR images is very complicated. Ship targets are susceptible to interference from complex backgrounds, and there are serious false detections and missed detections. (4) The ship targets in SAR images are characterized by a large aspect ratio, arbitrary direction and dense arrangement. Traditional horizontal box detection can cause non-target areas to interfere with the extraction of ship features, and it is difficult to accurately express the length, width and axial information of ship targets. To solve these problems, we propose an effective lightweight anchor-free detector called R-Centernet+ in the paper. Its features are as follows: the Convolutional Block Attention Module (CBAM) is introduced to the backbone network to improve the focusing ability on small ships; the Foreground Enhance Module (FEM) is used to introduce foreground information to reduce the interference of the complex background; the detection head that can output the ship angle map is designed to realize the rotation detection of ship targets. To verify the validity of the proposed model in this paper, experiments are performed on two public SAR image datasets, i.e., SAR Ship Detection Dataset (SSDD) and AIR-SARShip. The results show that the proposed R-Centernet+ detector can detect both inshore and offshore ships with higher accuracy than traditional models with an average precision of 95.11% on SSDD and 84.89% on AIR-SARShip, and the detection speed is quite fast with 33 frames per second.