Edge Loss Research Articles

SSCNet: A Spectrum-Space Collaborative Network for Semantic Segmentation of Remote Sensing Images

Semantic segmentation plays a pivotal role in the intelligent interpretation of remote sensing images (RSIs). However, conventional methods predominantly focus on learning representations within the spatial domain, often resulting in suboptimal discriminative capabilities. Given the intrinsic spectral characteristics of RSIs, it becomes imperative to enhance the discriminative potential of these representations by integrating spectral context alongside spatial information. In this paper, we introduce the spectrum-space collaborative network (SSCNet), which is designed to capture both spectral and spatial dependencies, thereby elevating the quality of semantic segmentation in RSIs. Our innovative approach features a joint spectral–spatial attention module (JSSA) that concurrently employs spectral attention (SpeA) and spatial attention (SpaA). Instead of feature-level aggregation, we propose the fusion of attention maps to gather spectral and spatial contexts from their respective branches. Within SpeA, we calculate the position-wise spectral similarity using the complex spectral Euclidean distance (CSED) of the real and imaginary components of projected feature maps in the frequency domain. To comprehensively calculate both spectral and spatial losses, we introduce edge loss, Dice loss, and cross-entropy loss, subsequently merging them with appropriate weighting. Extensive experiments on the ISPRS Potsdam and LoveDA datasets underscore SSCNet’s superior performance compared with several state-of-the-art methods. Furthermore, an ablation study confirms the efficacy of SpeA.

A densely connected LDCT image denoising network based on dual-edge extraction and multi-scale attention under compound loss.

Low dose computed tomography (LDCT) uses lower radiation dose, but the reconstructed images contain higher noise that can have negative impact in disease diagnosis. Although deep learning with the edge extraction operators reserves edge information well, only applying the edge extraction operators to input LDCT images does not yield overall satisfactory results. To improve LDCT images quality, this study proposes and tests a dual edge extraction multi-scale attention mechanism convolution neural network (DEMACNN) based on a compound loss. The network uses edge extraction operators to extract edge information from both the input images and the feature maps in the network, improving the utilization of the edge operators and retaining the images edge information. The feature enhancement block is constructed by fusing the attention mechanism and multi-scale module, enhancing effective information, while suppressing useless information. The residual learning method is used to learn the network, improving the performance of the network, and solving the problem of gradient disappearance. Except for the network structure, a compound loss function, which consists of the MSE loss, the proposed joint total variation loss, and the edge loss, is proposed to enhance the denoising ability of the network and reserve the edge of images. Compared with other advanced methods (REDCNN, CT-former and EDCNN), the proposed new network achieves the best PSNR and SSIM values in LDCT images of the abdomen, which are 33.3486 and 0.9104, respectively. In addition, the new network also performs well on head and chest image data. The experimental results demonstrate that the proposed new network structure and denoising algorithm not only effectively removes the noise in LDCT images, but also protects the edges and details of the images well.

FSSDD: Few‐shot steel defect detection based on multi‐scale semantic enhancement representation and mask category information mapping

AbstractSteel defect detection is important for industry production as it is tied to the product quality and production efficiency. However, previous steel defect detection methods based on deep convolutional neural networks heavily rely on large‐scale data for training and tend to have poor generalization ability for a novel defect category. In this paper, a novel few‐shot steel defect detection model based on multi‐scale semantic enhancement representation and mask category information mapping is introduced, where only a few annotated samples are acquired for the novel defect category. More concretely, three main components are built: an information‐guidance enhanced multi‐head detector is proposed to improve the representation of information in meta‐feature maps, a mask category representation module is designed to enhance the category feature representation of the mask region in the support set, and a novel multi‐scale category edge loss function is designed to assist the generation of category reweighting vector. Extensive experiments on the North‐east University few‐shot steel defect data set demonstrate that the proposed method significantly outperforms the state‐of‐the‐art methods and verify its effectiveness through ablation studies.

Multi-Domain Rapid Enhancement Networks for Underwater Images

Images captured during marine engineering operations suffer from color distortion and low contrast. Underwater image enhancement helps to alleviate these problems. Many deep learning models can infer multi-source data, where images with different perspectives exist from multiple sources. To this end, we propose a multichannel deep convolutional neural network (MDCNN) linked to a VGG that can target multi-source (multi-domain) underwater image enhancement. The designed MDCNN feeds data from different domains into separate channels and implements parameters by linking VGGs, which improves the domain adaptation of the model. In addition, to optimize performance, multi-domain image perception loss functions, multilabel soft edge loss for specific image enhancement tasks, pixel-level loss, and external monitoring loss for edge sharpness preprocessing are proposed. These loss functions are set to effectively enhance the structural and textural similarity of underwater images. A series of qualitative and quantitative experiments demonstrate that our model is superior to the state-of-the-art Shallow UWnet in terms of UIQM, and the performance evaluation conducted on different datasets increased by 0.11 on average.

A High-Precision Ensemble Model for Forest Fire Detection in Large and Small Targets

Forest fires are major forestry disasters that cause loss of forest resources, forest ecosystem safety, and personal injury. It is often difficult for current forest fire detection models to achieve high detection accuracy on both large and small targets at the same time. In addition, most of the existing forest fire detection models are single detection models, and using only a single model for fire detection in a complex forest environment has a high misclassification rate, and the accuracy rate needs to be improved. Aiming at the above problems, this paper designs two forest fire detection models (named WSB and WSS) and proposes an integrated learning-based forest fire detection model (named WSB_WSS), which also obtains high accuracy in the detection of forest fires with large and small targets. In order to help the model predict the location and size of forest fire targets more accurately, a new edge loss function, Wise-Faster Intersection over Union (WFIoU), is designed in this paper, which effectively improves the performance of the forest fire detection algorithm. The WSB model introduces the Simple-Attention-Module (SimAM) attention mechanism to make the image feature extraction more accurate and introduces the bi-directional connectivity and cross-layer feature fusion to enhance the information mobility and feature expression ability of the feature pyramid network. The WSS model introduces the Squeeze-and-Excitation Networks (SE) attention mechanism so that the model can pay more attention to the most informative forest fire features and suppress unimportant features, and proposes Spatial Pyramid Pooling-Fast Cross Stage Partial Networks (SPPFCSPC) to enable the network to extract features better and speed up the operation of the model. The experimental findings demonstrate that the WSB model outperforms other approaches in the context of identifying forest fires characterized by small-scale targets, achieving a commendable accuracy rate of 82.4%, while the WSS model obtains a higher accuracy of 92.8% in the identification of large target forest fires. Therefore, in this paper, a more efficient forest fire detection model, WSB_WSS, is proposed by integrating the two models through the method of Weighted Boxes Fusion (WBF), and the accuracy of detecting forest fires characterized by small-scale targets attains 83.3%, while for forest fires with larger dimensions, the accuracy reaches an impressive 93.5%. This outcome effectively leverages the strengths inherent in both models, consequently achieving the dual objective of high-precision detection for both small and large target forest fires concurrently.

Enhanced seismic data segmentation using an assembled scSE-Res-UNet deep neural network

To enhance the accuracy and robustness of seismic data segmentation, we propose a novel deep-learning network, named scSE-Res-UNet. We select a U-Net as the backbone and augment it with residual blocks to form a Res-UNet, enabling multi-scale feature extraction while reducing edge loss between strata. Furthermore, we introduce Spatial and Channel Squeeze & Excitation (scSE) attention blocks in the decoder to selectively emphasize meaningful features and long-range dependencies in the data. The scSE blocks suppress unimportant features and noise while enhancing spatial context. We apply the proposed scSE-Res-UNet to a public F3 seismic dataset and compare performance against a CNN benchmark model. Experiments demonstrate the proposed architecture achieves superior data segmentation accuracy and robustness. The integration of residual learning and scSE attention mechanisms enhances the network's capacity to extract geologically meaningful representations from the seismic data. This leads to improved delineation of stratigraphic units with similar characters that are challenging to differentiate. The proposed scSE-Res-UNet effectively combines proven deep learning techniques to advance seismic interpretation.

Ink painting style transfer using asymmetric cycle-consistent GAN

Chinese ink painting, an artistic and cultural treasure, necessitates automatic generation for its preservation and evolution. We've innovatively observed and validated that the domain information between photographs and ink paintings, is asymmetrical, which has been overlooked by current style transfer algorithms. We propose an innovative generative adversarial network featuring an asymmetric cyclic consistency structure to address this in ink painting style transfer. This structure uses generators of differing capabilities to align with the asymmetry in transformation directions, improving image quality and model optimization speed. Additionally, we introduce two unique loss functions within the network. The salient edge loss intensifies the subject in the real photo and enhances the edge stroke of the drawn subject, a distinct attribute of ink painting. The feature-wise cycle consistency loss is designed to speed up model optimization. We've also built a Chinese bird ink painting dataset to validate effectiveness of the model. Extensive experiments on this and a public dataset demonstrate that our algorithm can comprehensively learn various stylistic features of ink painting, especially regarding brushstroke style, ink diffusion, and detail preservation. Furthermore, the quantitative results indicate our approach achieves superior results in generation quality and model efficiency compared to existing methods. For instance, compared to the most recent style transfer method, our method achieves an average decrease of 9.44% and 25.32% for FID and KID metrics across three datasets and reduces training and inference time costs by 46.81% and 22.71% respectively.

DSRSS-Net: Improved-Resolution Snow Cover Mapping from FY-4A Satellite Images Using the Dual-Branch Super-Resolution Semantic Segmentation Network

The Qinghai–Tibet Plateau is one of the regions with the highest snow accumulation in China. Although the Fengyun-4A (FY4A) satellite is capable of monitoring snow-covered areas in real time and on a wide scale at high temporal resolution, its spatial resolution is low. In this study, the Qinghai–Tibet Plateau, which has a harsh climate with few meteorological stations, was selected as the study area. We propose a deep learning model called the Dual-Branch Super-Resolution Semantic Segmentation Network (DSRSS-Net), in which one branch focuses with super resolution to obtain high-resolution snow distributions and the other branch carries out semantic segmentation to achieve accurate snow recognition. An edge enhancement module and coordinated attention mechanism were introduced into the network to improve the classification performance and edge segmentation effect for cloud versus snow. Multi-task loss is also used for optimization, including feature affinity loss and edge loss, to obtain fine structural information and improve edge segmentation. The 1 km resolution image obtained by coupling bands 1, 2, and 3; the 2 km resolution image obtained by coupling bands 4, 5, and 6; and the 500 m resolution image for a single channel, band 2, were inputted into the model for training. The accuracy of this model was verified using ground-based meteorological station data. Snow classification accuracy, false detection rate, and total classification accuracy were compared with the MOD10A1 snow product. The results show that, compared with MOD10A1, the snow classification accuracy and the average total accuracy of DSRSS-Net improved by 4.45% and 5.1%, respectively. The proposed method effectively reduces the misidentification of clouds and snow, has higher classification accuracy, and effectively improves the spatial resolution of FY-4A satellite snow cover products.

Quantitative Analysis of the Interface between Titanium Dioxide Support and Noble Metal by Electron Energy Loss Spectroscopy.

The interface structure of supported catalysts plays a significant role in the strong metal-support interactions (SMSI). However, it remains limited on interpreting interface structures, thus affecting the understanding of SMSI origin and impact on catalytic performance. Herein, electronic energy loss spectroscopy was adopted to characterize the interface microstructures of Pt/TiO2 materials. After high-temperature reduction processing, it was observed that the coating on the surface of the Pt metal particles was TiOx. Then, based on Gaussian function fitting, an effective and valid method was established for quantitative analysis on Ti L edge loss spectrum. This method allowed us to accurately determine the stoichiometric number of TiOx phases. In order to probe the classical phenomenon of strong metal-support interactions in more detail, we also discussed and analyzed the origin of TiOx and its effect on the electronic structure of the material using density functional theory calculations. The structure of surfaces and interfaces as well as the chemical evolution of supported catalysts on a microscale have been revealed, thereby providing a new analysis method and research perspectives for the future study of metal-support interactions.

Wavelet transform and edge loss-based three-stage segmentation model for retinal vessel

Retinal vessel segmentation is a rapid method for the diagnosis of ocular diseases. By applying deep learning-based techniques to retinal images, more structural information about retinal vessels can be extracted to accurately assess the extent and classification of ocular diseases. However, current segmentation networks typically consist of a single network, making them vulnerable to noise, decreased image quality, and other interfering factors, resulting in erroneous segmentation outcomes. Additionally, the traditional skip connection mechanism introduces noise from the encoder features into the decoder, which reduces the refinement of the final segmentation result. A three-stage fundus vessel segmentation model called EWSNet is proposed to address these issues. The EWSNet utilizes two different models to extract and reconstruct coarse and fine blood vessels, respectively. The reconstructed results are fed into the refinement network to rebuild the edge portion of the retinal vessels, achieving higher segmentation performance. Within the framework of EWSNet, a wavelet-transformation-based sampling module is used to effectively suppress high-frequency noise in the features while using low-frequency features to reconstruct vascular information. Besides, a new edge loss function (E-BCE Loss) is designed to encourage more precise predictions at the segmentation edges. Experimental results on CHASE_DB1, HRF, STARE, and a newly collected ultra-wide-angle fundus dataset (UWF) demonstrate that EWSNet has more robust segmentation performance in the microvascular region compared to the current mainstream models. The code is available at: https://github.com/xuecheng990531/EWSNet.

SU2GE-Net: a saliency-based approach for non-specific class foreground segmentation

Salient object detection is vital for non-specific class subject segmentation in computer vision applications. However, accurately segmenting foreground subjects with complex backgrounds and intricate boundaries remains a challenge for existing methods. To address these limitations, our study proposes SU2GE-Net, which introduces several novel improvements. We replace the traditional CNN-based backbone with the transformer-based Swin-TransformerV2, known for its effectiveness in capturing long-range dependencies and rich contextual information. To tackle under and over-attention phenomena, we introduce Gated Channel Transformation (GCT). Furthermore, we adopted an edge-based loss (Edge Loss) for network training to capture spatial-wise structural details. Additionally, we propose Training-only Augmentation Loss (TTA Loss) to enhance spatial stability using augmented data. Our method is evaluated using six common datasets, achieving an impressive F_{beta } score of 0.883 on DUTS-TE. Compared with other models, SU2GE-Net demonstrates excellent performance in various segmentation scenarios.

P-TransUNet: an improved parallel network for medical image segmentation

Deep learning-based medical image segmentation has made great progress over the past decades. Scholars have proposed many novel transformer-based segmentation networks to solve the problems of building long-range dependencies and global context connections in convolutional neural networks (CNNs). However, these methods usually replace the CNN-based blocks with improved transformer-based structures, which leads to the lack of local feature extraction ability, and these structures require a huge number of data for training. Moreover, those methods did not pay attention to edge information, which is essential in medical image segmentation. To address these problems, we proposed a new network structure, called P-TransUNet. This network structure combines the designed efficient P-Transformer and the fusion module, which extract distance-related long-range dependencies and local information respectively and produce the fused features. Besides, we introduced edge loss into training to focus the attention of the network on the edge of the lesion area to improve segmentation performance. Extensive experiments across four tasks of medical image segmentation demonstrated the effectiveness of P-TransUNet, and showed that our network outperforms other state-of-the-art methods.

An Improved Combination of Faster R-CNN and U-Net Network for Accurate Multi-Modality Whole Heart Segmentation.

Detailed information of substructures of the whole heart is usually vital in the diagnosis of cardiovascular diseases and in 3D modeling of the heart. Deep convolutional neural networks have been demonstrated to achieve state-ofthe-art performance in 3D cardiac structures segmentation. However, when dealing with high-resolution 3D data, current methods employing tiling strategies usually degrade segmentation performances due to GPU memory constraints. This work develops a two-stage multi-modality whole heart segmentation strategy, which adopts an improved Combination of Faster R-CNN and 3D U-Net (CFUN+). More specifically, the bounding box of the heart is first detected by Faster R-CNN, and then the original Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) images of the heart aligned with the bounding box are input into 3D U-Net for segmentation. The proposed CFUN+ method redefines the bounding box loss function by replacing the previous Intersection over Union (IoU) loss with Complete Intersection over Union (CIoU) loss. Meanwhile, the integration of the edge loss makes the segmentation results more accurate, and also improves the convergence speed. The proposed method achieves an average Dice score of 91.1% on the Multi-Modality Whole Heart Segmentation (MM-WHS) 2017 challenge CT dataset, which is 5.2% higher than the baseline CFUN model, and achieves state-of-the-art segmentation results. In addition, the segmentation speed of a single heart has been dramatically improved from a few minutes to less than 6 seconds.

In pursuit of gold standard shade guide method, a comparative in-vivo study between manual and digital shade guides

The available shade guide system in dentistry includes Manual and Digital. Although there is lot of research going on better system to choose dental shade but still there are many factors including source of light illuminating the tooth, the characteristics of tooth and the observer viewing the tooth which needs to study further. In this study, a comparative analysis is made between the Manual and Digital shade guide systems.: The objective of the study is to record the manual (subjective method) and a digital shade (Objective method) of 500 subjects by two observers (Prosthodontist 1 {P1} and Prosthodontist 2 {P2}). : 500 patients having a complete set of anterior teeth with sound maxillary central incisors were selected for the study. Manual shade selection (Vita 3D Master) of 50% of subjects is made by P1. Then P2 does shade selection of the same subjects by digital method (Vita Easy Shade Advance) and records the deviation of values including ∆E, ∆L, ∆C, and ∆h by digital shade guide from the manual shade as recorded by P1 using verification mode of Digital shade guide. Recording of shades and deviation values is done for the remaining 50 % of patients by role reversal between P1 and P2.: The kappa correlation values for Hue value and chroma when P1 records manually and P2 documents shade by digital shade guide include 0.26, 0.42, and 0.13 indicating fair, moderate, and slight to no agreement respectively. The kappa correlation values for Hue value and chroma when P2 records manually and P1 records shade by digital shade guide include -0.08, 0.51 and 0.12 indicating no agreement, moderate and slight to no agreement respectively.It is advisable to use technology to overcome the errors of subjectivity. Due to edge loss with a spectrophotometer, especially in the incisal and cervical region, a combination of digital and manual systems always comes in handy when deciding on a particular shade or shade map.

Multiscale Attention Fusion for Depth Map Super-Resolution Generative Adversarial Networks.

Color images have long been used as an important supplementary information to guide the super-resolution of depth maps. However, how to quantitatively measure the guiding effect of color images on depth maps has always been a neglected issue. To solve this problem, inspired by the recent excellent results achieved in color image super-resolution by generative adversarial networks, we propose a depth map super-resolution framework with generative adversarial networks using multiscale attention fusion. Fusion of the color features and depth features at the same scale under the hierarchical fusion attention module effectively measure the guiding effect of the color image on the depth map. The fusion of joint color-depth features at different scales balances the impact of different scale features on the super-resolution of the depth map. The loss function of a generator composed of content loss, adversarial loss, and edge loss helps restore clearer edges of the depth map. Experimental results on different types of benchmark depth map datasets show that the proposed multiscale attention fusion based depth map super-resolution framework has significant subjective and objective improvements over the latest algorithms, verifying the validity and generalization ability of the model.

An adaptive method of measuring the rake face wear of end mills based on image feature point set registration

An acceptable tool wear evaluation and acquisition approach is critical for developing an accurate, rapid, and practical online tool wear prediction system. The one-dimensional crater wear depth is difficult to quantify, and relying solely on flank wear width is very simplistic. External conditions such as vibration and illumination have a large impact on two-dimensional wear area and three-dimensional volume. In practice, these are difficult to successfully characterize tool wear. This work proposes a two-dimensional rake face edge loss area as an additional indicator to assess milling cutter wear. An adaptive measurement approach based on 2D image feature point set matching is presented to suppress the area calculation error caused by image position offset and chips or built-up edges. 201 images were collected in two groups of cutting experiments for comparison with the proposed method and the other two methods, manual registration and non-registration. The results show that the proposed method's measurement values are closer to the actual values and have a higher measurement efficiency. Furthermore, the proposed indicator is sensitive to micro-scale changes to the blade edge, especially at the tip, and can be used as a supplement one to monitor tool wear more comprehensively.

A comparative study of mean-line models based on enthalpy loss and analysis of a cavity-structured radial turbine for solar hybrid microturbine applications

Lightweight components are essential for the future of the energy and aerospace fields. This paper presents an innovative radial turbine design that utilizes a cavity structure on the rotor for solar hybrid microturbines. The performance of the cavity-structured radial turbine is studied using mean-line models, computational fluid dynamics (CFD), and finite element analysis (FEA) simulations. The radial turbine is compared using air and supercritical carbon dioxide (sCO2) as the working fluid. The mean-line models based on enthalpy losses from recent literature are validated using commercial code Rital and CFD results. The impact of the isentropic velocity ratio (νs) and absolute flow angle at the rotor inlet (α1) on the optimum design for the rotor radial turbine is also demonstrated. A comparative loss enthalpy breakdown on the rotor is examined, such as incidence loss, passage loss, trailing edge loss, clearance loss, exit loss, and windage loss. The results show that the recent mean-line models based on enthalpy losses can accurately estimate the stage efficiency for both air and sCO2 turbines. By comparing the simulation results obtained from 3D-CFD and mean-line models, the efficiency prediction deviation is within an acceptable range of tolerance for both turbines when comparing the results obtained from 3D-CFD and mean-line models. The FE analysis reveals that the maximum stress decreased significantly for the air and sCO2 turbine after the rotor blade was added using the cavity structure inside. The maximum stress value on the rotor with the cavity structure design is still below the yield strength of the material limits. Furthermore, after the variations of the cavity structure design were added to the rotor blade of both turbines, the results were promising when the dangerous resonance at below 4EO (Engine Order) was minimized. The results of this study provide a promising direction for the development of high-efficiency and lightweight turbines in the energy and aerospace fields.

SDAN-MD: Supervised dual attention network for multi-stage motion deblurring in frontal-viewing vehicle-camera images

Motion blur in images usually distorts the information of objects, thus degrading the performance of semantic segmentation. However, there is no previous research on improving the segmentation performance by restoring the frontal-viewing vehicle-camera images taken under motion blur. Therefore, this study proposes a supervised dual attention network for multi-stage motion deblurring (SDAN-MD) for this task. In SDAN-MD, a supervised dual attention module (SDAM) is proposed, which adopts the supervised spatial and channel attention mechanisms to provide a supervisory signal of ground truth. In addition to Charbonnier loss and edge loss, we use perceptual loss utilizing Euclidean distance based on feature maps obtained from the segmentation network.Experiments were conducted with the motion blurred databases from the two open databases of road scene, Cambridge driving Labeled Video Database (CamVid) and Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago (KITTI).The results show that the proposed SDAN-MD achieves 92.89% and 87.27% pixel accuracies in semantic segmentation using these two databases, respectively, outperforming the state-of-the-art methods.

TP-Net: Two-Path Network for Retinal Vessel Segmentation.

Refined and automatic retinal vessel segmentation is crucial for computer-aided early diagnosis of retinopathy. However, existing methods often suffer from mis-segmentation when dealing with thin and low-contrast vessels. In this paper, a two-path retinal vessel segmentation network is proposed, namely TP-Net, which consists of three core parts, i.e. main-path, sub-path, and multi-scale feature aggregation module (MFAM). Main-path is to detect the trunk area of the retinal vessels, and the sub-path to effectively capture edge information of the retinal vessels. The prediction results of the two paths are combined by MFAM, obtaining refined segmentation of retinal vessels. In the main-path, a three-layer lightweight backbone network is elaborately designed according to the characteristics of retinal vessels, and then a global feature selection mechanism (GFSM) is proposed, which can autonomously select features that are more important for the segmentation task from the features at different layers of the network, thereby, enhancing the segmentation capability for low-contrast vessels. In the sub-path, an edge feature extraction method and an edge loss function are proposed, which can enhance the ability of the network to capture edge information and reduce the mis-segmentation of thin vessels. Finally, MFAM is proposed to fuse the prediction results of main-path and sub-path, which can remove background noises while preserving edge details, and thus, obtaining refined segmentation of retinal vessels. The proposed TP-Net has been evaluated on three public retinal vessel datasets, namely DRIVE, STARE, and CHASE DB1. The experimental results show that the TP-Net achieved a superior performance and generalization ability with fewer model parameters compared with the state-of-the-art methods.

Multi-Scale Feature Fusion Network for Low-Dose CT Denoising.

Computed tomography (CT) is an imaging technique extensively used in medical treatment, but too much radiation dose in a CT scan will cause harm to the human body. Decreasing the dose of radiation will result in increased noise and artifacts in the reconstructed image, blurring the internal tissue and edge details. To get high-quality CT images, we present a multi-scale feature fusion network (MSFLNet) for low-dose CT (LDCT) denoising. In our MSFLNet, we combined multiple feature extraction modules, effective noise reduction modules, and fusion modules constructed using the attention mechanism to construct a horizontally connected multi-scale structure as the overall architecture of the network, which is used to construct different levels of feature maps at all scales. We innovatively define a composite loss function composed of pixel-level loss based on MS-SSIM-L1 and edge-based edge loss for LDCT denoising. In short, our approach learns a rich set of features that combine contextual information from multiple scales while maintaining the spatial details of denoised CT images. Our laboratory results indicate that compared with the existing methods, the peak signal-to-noise ratio (PSNR) value of CT images of the AAPM dataset processed by the new model is 33.6490, and the structural similarity (SSIM) value is 0.9174, which also achieves good results on the Piglet dataset with different doses. The results also show that the method removes noise and artifacts while effectively preserving CT images' architecture and grain information.