Traditional Image Research Articles

The clinical application value of mixed reality in robotic laparoscopic partial nephrectomy

PurposeRobot-assisted laparoscopic partial nephrectomy (RAPN) has become a key technology in the treatment of renal tumors. Effective preoperative planning and precise intraoperative navigation are critical to a successful surgical outcome. This study aimed to evaluate the clinical application value of mixed reality (MR) in robotic nephron-sparing partial nephrectomy for patients with renal tumors of different complexity based on the R.E.N.A.L. score.Patients and methodsA retrospective analysis was conducted on 68 eligible patients with renal cancer who underwent RAPN at The Second Affiliated Hospital of Nanchang University from January 2021 to December 2023, with postoperative pathology confirmation. Patients were divided into two groups: the MR group, with 28 cases, and the traditional imaging (control) group, with 40 cases. All patients underwent mid-abdominal CT plain scans and enhancements. The MR group utilized three-dimensional reconstruction of CT data and employed 3D tablets and HoloLens glasses for preoperative discussions, surgical planning, and intraoperative guidance. Collect clinical data and metrics to assess surgical outcomes, as well as evaluate performance in areas such as preoperative discussions, doctor-patient communication, surgical planning, and intraoperative navigation.ResultsCompared to robot-assisted partial nephrectomy in the control group, the MR group experienced a reduction in operation time by approximately 30 min [(135.89 ± 23.494) min vs. (165.00 ± 34.320) min, P< 0.001)] and a decrease in ischemia time by around 2.5 min [(20.36 ± 3.956) min vs. (23.80± 6.889) min, P = 0.02)]. Within the subgroup with a R.E.N.A.L. score of less than 7 points, the MR group only showed a significant reduction in operation time [(134.55 ± 150.190) min vs. (150.19 ± 28.638) min, P = 0.045], with no notable differences in other parameters. For the subgroup with a R.E.N.A.L. score of 7 points or higher, the MR group exhibited shorter operation time [(140.83 ± 25.183) min vs. (195.77 ± 23.080) min, P< 0.001] and reduced warm ischemia time [(21.17 ± 2.714) min vs. (28.85 ± 7.570) min, P = 0.029]. Additionally, there was less estimated blood loss [(53.33 ± 5.164) min vs. (114.62 ± 80.376) min, P = 0.018]. All patients had negative resection margins, indicating equivalent therapeutic outcomes between the two groups.ConclusionIn comparison to traditional RAPN, MR technology demonstrates the ability to decrease operation time and warm ischemia time all the while maintaining equivalent curative outcomes. Additionally, it enhances preoperative discussions, doctor-patient interactions, preoperative strategizing, and intraoperative navigation, particularly excelling in complex renal tumor cases of RAPN, where its benefits are most pronounced.

Improved visualisation of ACP-engineered osteoblastic spheroids: a comparative study of contrast-enhanced micro-CT and traditional imaging techniques

This study investigates osteoblastic cell spheroid cultivation methods, exploring flat-bottom, U-bottom, and rotary flask techniques with and without amorphous calcium phosphate (ACP) supplementation to replicate the 3D bone tissue microenvironment. ACP particles derived from eggshell waste exhibit enhanced osteogenic activity in 3D models. However, representative imaging of intricate 3D tissue-engineered constructs poses challenges in conventional imaging techniques due to notable scattering and absorption effects in light microscopy, and hence limited penetration depth. We investigated contrast-enhanced micro-CT as a methodological approach for comprehensive morphological 3D-analysis of thein-vitromodel and compared the technique with confocal laser scanning microscopy, scanning electron microscopy and classical histology. Phosphotungstic acid and iodine-based contrast agents were employed for micro-CT imaging in laboratory and synchrotron micro-CT imaging. Results revealed spheroid shape variations and structural integrity influenced by cultivation methods and ACP particles. The study underscores the advantage of 3D spheroid models over traditional 2D cultures in mimicking bone tissue architecture and cellular interactions, emphasising the growing demand for novel imaging techniques to visualise 3D tissue-engineered models. Contrast-enhanced micro-CT emerges as a promising non-invasive imaging method for tissue-engineered constructs containing ACP particles, offering insights into sample morphology, enabling virtual histology before further analysis.

Thumbnail-Preserving Encryption Based on Binary–Decimal Hybrid and a Generalized Coupled Logistic Map

Traditional image encryption methods typically make a trade-off between privacy protection and visual usability. To address this issue, a novel thumbnail-preserving encryption scheme is proposed based on a hybrid of decimal and binary pixel values in three pixel groups, utilizing the excellent dynamic characteristics of a generalized coupled logistic map. Theoretically, it is proved that the core algorithm of this scheme is sum-preserving. Experimentally, it is demonstrated that this scheme not only excels in efficiency and resistance to attacks, effectively safeguarding image privacy, but also retains high-quality thumbnails after encryption, achieving both objectives of image privacy protection and visual usability.

Scalable deep learning artificial intelligence histopathology slide analysis and validation.

Deep learning involves an artificial intelligence (AI) approach and has been shown to provide superior performance for automating image recognition tasks, as well as exceeding human capabilities in both time and accuracy. Histopathology diagnostics is one of the more popular challenges at the intersection of artificial intelligence, computer vision, and medicine. Developing methods to automatically detect and identify pathologies in digitized histology slides imposes unique challenges due to the large size of these images and the complexity of the features present in biological tissue. Most methods that are capable of human-level recognition in histopathology are tuned to a specific problem since the computational complexity exceeds that of traditional image classification problems. In the current study, a deep learning approach is developed and presented that can be trained to locate and accurately classify different types of pathologies in gigapixel digitized histology slides along with completing the binary disease classification for the entire image. The approach uses a novel pyramid tiling approach to take advantage of spatial awareness around the area to be classified, while maintaining efficiency and scalability for gigapixel images. The approach is trained and validated on a wide variety of tissue types (i.e., testis, ovary, prostate, kidney) and pathologies taken from an epigenetically altered histology study at Washington State University. The newly developed procedure was optimized and validated along with comparison and validation on public histology datasets. The current developed procedure was found to be optimal and more reproducible when compared to manual procedures, and optimal to previous protocols that used fragmented tissue or slide analysis. Observations demonstrate that the deep learning histopathology analysis is significantly more efficient and accurate than standard manual histopathology analysis.

A New Encryption Algorithm Utilizing DNA Subsequence Operations for Color Images

The computer network has fundamentally transformed modern interactions, enabling the effortless transmission of multimedia data. However, the openness of these networks necessitates heightened attention to the security and confidentiality of multimedia content. Digital images, being a crucial component of multimedia communications, require robust protection measures, as their security has become a global concern. Traditional color image encryption/decryption algorithms, such as DES, IDEA, and AES, are unsuitable for image encryption due to the diverse storage formats of images, highlighting the urgent need for innovative encryption techniques. Chaos-based cryptosystems have emerged as a prominent research focus due to their properties of randomness, high sensitivity to initial conditions, and unpredictability. These algorithms typically operate in two phases: shuffling and replacement. During the shuffling phase, the positions of the pixels are altered using chaotic sequences or matrix transformations, which are simple to implement and enhance encryption. However, since only the pixel positions are modified and not the pixel values, the encrypted image’s histogram remains identical to the original, making it vulnerable to statistical attacks. In the replacement phase, chaotic sequences alter the pixel values. This research introduces a novel encryption technique for color images (RGB type) based on DNA subsequence operations to secure these images, which often contain critical information, from potential cyber-attacks. The suggested method includes two main components: a high-speed permutation process and adaptive diffusion. When implemented in the MATLAB software environment, the approach yielded promising results, such as NPCR values exceeding 98.9% and UACI values at around 32.9%, demonstrating its effectiveness in key cryptographic parameters. Security analyses, including histograms and Chi-square tests, were initially conducted, with passing Chi-square test outcomes for all channels; the correlation coefficient between adjacent pixels was also calculated. Additionally, entropy values were computed, achieving a minimum entropy of 7.0, indicating a high level of randomness. The method was tested on specific images, such as all-black and all-white images, and evaluated for resistance to noise and occlusion attacks. Finally, a comparison of the proposed algorithm’s NPCR and UAC values with those of existing methods demonstrated its superior performance and suitability.

TIA-UNet: Transformer-Enhanced Deep Learning for Adolescent Idiopathic Scoliosis Spinal X-Ray Image Segmentation

Abstract Adolescent Idiopathic Scoliosis (AIS) is a common spinal deformity where precise diagnosis is crucial for developing effective treatment strategies. Traditional manual X-ray image analysis is time-consuming and highly dependent on the operator’s expertise, thus constraining diagnostic efficiency and accuracy.This study aimed to develop an automated thoracolumbar spine segmentation method utilizing deep learning to enhance the efficiency and accuracy of AIS diagnosis. We introduced TIA-UNet, an innovative network architecture that combines Convolutional Neural Networks (CNN) with Transformer models.By integrating the IR Block and DA Block, TIA-UNet was optimized for feature extraction and multi-scale information fusion. The model underwent training and validation using our established Adolescent Scoliosis Medical Dataset (ASMD).TIA-UNet attained a Dice similarity coefficient of 90.02%, a Mean Intersection over Union (MIoU) of 81.96%, and a Hausdorff Distance (HD) of 4.09, significantly surpassing current state-of-the-art methods such as UNet and TransUNet. Moreover, TIA-UNet exhibited superior computational efficiency regarding parameter count, inference time, and floatingpoint operations (FLOPs).As an automated medical image segmentation algorithm, TIA-UNet enhanced segmentation accuracy while maintaining high computational efficiency, demonstrating significant potential for clinical diagnostic applications. This study provides compelling evidence supporting the utilization of deep learning
techniques in medical image analysis.

Machine vision-based detection of surface defects in cylindrical battery cases

Automotive 21700 series lithium batteries are prone to surface defects during production and transportation, thus affecting their performance, so we propose a full-surface defect detection method for battery cases based on the synthesis of traditional image processing and deep learning to address this problem. First, the mechanism of surface defects on a battery case is analysed, and the types of surface defects are summarized. A suitable platform for image acquisition is designed for the severely reflective surface. Since there is no publicly available defect dataset for cylindrical battery cases, a defect dataset is established, and the dataset is augmented and expanded via the traditional method and the ACGAN model. For the defects on the top of the battery case, a traditional image processing algorithm is used to combine the roundness and the area of the area pixels to make a comprehensive judgement. For the defects on the bottom and side of the battery case, after comparing and analysing a variety of deep learning networks, the YOLOv7 model is selected to address the data characteristics of the large experimental inputs and the small targeted defects. To improve the accuracy of the model to meet the needs of real-time detection in industry, the CA attention mechanism, the DYHEAD dynamic detector head, and the slicing-assisted super inference (SAHI) method are used, with a final map accuracy reaching 98.1 %, which is 2.7 % higher than that of the initial model. Compared with mainstream target detection algorithms, the algorithm in this paper has good detection performance for cylindrical battery case defect detection and can be better applied to real-time detection in industry.

Contribution of 3D virtual modeling in locating hepatic metastases, particularly “vanishing tumors”: a pilot study

Aim: This article aims to explore how three dimensions (3D) virtual modeling enhances accuracy and efficiency in detecting hepatic metastases, with a specific focus on “vanishing tumors” that are difficult to detect using traditional imaging techniques. It also aims to demonstrate the potential impact of these advanced technologies on improving diagnostic and treatment strategies for patients with liver cancer. Methods: Eight patients with liver metastases from colorectal cancer were studied using magnetic resonance imaging (MRI) and 3D virtual modeling to enhance surgical planning by accurately locating lesions. The concordance between these imaging techniques and the Gold Standard was assessed using Gwet’s AC1 coefficient, with statistical analysis performed through resampling methods and non-parametric tests due to non-normal AC1 distribution. Liver segmentation was conducted semi-automatically using SurgiPrint software, which is crucial for detecting lesions undetectable post-chemotherapy. The study’s evaluation model involved questionnaires for medical professionals across four cohorts, aiming to determine the 3D model’s effectiveness in identifying lesion locations for surgery. Results: This study investigated the efficacy of 3D virtual modeling in identifying hepatic metastases, particularly comparing its accuracy with traditional MRI in locating lesions. The findings indicate that MRI generally provided better concordance with the Gold Standard for lesion localization, except for a few experienced users who had prior familiarity with the 3D models. Despite the mixed results in accuracy, the study suggests potential benefits of 3D modeling in enhancing surgical planning and execution, particularly in detecting “vanishing” liver metastases that are difficult to visualize with standard imaging techniques. The research aligns with broader evidence indicating the utility of 3D models in improving outcomes in hepatic surgery by enabling more precise resections and reducing postoperative complications. However, the study also notes the challenge of quantifying the added value of 3D modeling due to the unique nature of each surgical case and the potential bias in the user experience with the technology. Conclusion: The conclusion of this study underscores the significant potential of 3D virtual modeling in enhancing the precision of locating and resecting hepatic metastases, particularly those elusive “vanishing tumors”. This research underscores the value of integrating advanced 3D imaging techniques into surgical practice, suggesting a paradigm shift toward more accurate and less invasive liver surgeries. Future studies should aim to further quantify the benefits of 3D modeling in surgical outcomes, investigate its utility across different surgical experiences, and explore the integration of artificial intelligence (AI) to maximize its effectiveness in clinical settings.

ENHANCED TRANSFORMER-BASED DEEP KERNEL FUSED SELF ATTENTION MODEL FOR LUNG NODULE SEGMENTATION AND CLASSIFICATION

Lung cancer remains a leading cause of mortality globally, with outcomes heavily dependent on the timeliness and accuracy of diagnosis. Traditional medical imaging techniques, while foundational in detecting lung nodules, often falter in distinguishing malignant from benign lesions with high precision, largely due to their inability to contextualize the complex spatial relationships within the images. Precise segmentation and classification of lung nodules is crucial for the early detection of lung cancer. This paper presents a novel deep learning model that incorporates a transformer block to improve the performance of lung cancer detection and classification. From performance evaluation, it is evident that our proposed model has an average accuracy of 93%. 05%, which is superior to the existing D3DR _ MKCA model with a mean accuracy of 91.53%. These findings are especially important for the identification of Adenocarcinoma and Small Cell Carcinoma, as improvements in the precision and recall factors have been achieved in these cases.

Research on the Laser Scattering Characteristics of Three-Dimensional Imaging Based on Electro–Optical Crystal Modulation

In this paper, we construct a laser 3D imaging simulation model based on the 3D imaging principle of electro–optical crystal modulation. Unlike the traditional 3D imaging simulation method, this paper focuses on the laser scattering characteristics of the target scene. To accurately analyze and simulate the scattering characteristic model of the target under laser irradiation, we propose a BRDF (Bidirectional Reflectance Distribution Function) model fitting algorithm based on the hybrid BBO–Firefly model, which can accurately simulate the laser scattering distribution of the target at different angles. Finally, according to the fitted scattering characteristic model, we inverted the target imaging gray map. We used the laser 3D imaging restoration principle to reconstruct the 3D point cloud of the target to realize the laser 3D imaging of the target.

Automatic Segmentation Model and Parameter Extraction Algorithm for Lightning Whistlers

AbstractBased on the magnetic field data recorded by the ZH‐1 electromagnetic satellite, we cerat a training set of 1,300 spectrograms containing the dispersion spectrum of lightning whistlers (LW). The Segment Anything Model (SAM) in the field of image segmentation is trained through the training set to obtain a fine‐tuned SAM model that can be used to detect and segment the dispersion spectrum of LW at pixel level. All track regions of LW are effectively separated from other non‐lightning whistlers regions in the spectrograms after being segmented by the model. The segmentation effect is excellent and detection accuracy is 96.89%, which is better than the previous segmentation model for LW based on ground station data. Then we apply the traditional image processing methods to extract the dispersion spectrum of LW one by one, and develop an algorithm to automatically extract the physical parameters of each LW. The root mean square error between the automatically extracted dispersion parameter and the manually extracted dispersion parameter is only 0.1654 s1/2. The model and algorithm studied in this paper are employed to analyze the dispersion of LW received by the ZH‐1 satellite over China. It is found that the whistlers dispersion received by satellites during summer in the northern hemisphere and summer in the southern hemisphere shows opposite trends with receiving latitude. Both trends can be explained by the relationship between the dispersion and the length of propagation paths of LW.

A Knowledge Base Driven Task-Oriented Image Semantic Communication Scheme

With the development of artificial intelligence and computer hardware, semantic communication has been attracting great interest. As an emerging communication paradigm, semantic communication can reduce the requirement for channel bandwidth by extracting semantic information. This is an effective method that can be applied to image acquisition of unmanned aerial vehicles, which can transmit high-data-volume images within the constraints of limited available bandwidth. However, the existing semantic communication schemes fail to adequately incorporate the guidance of task requirements into the semantic communication process and are difficult to adapt to the dynamic changes of tasks. A task-oriented image semantic communication scheme driven by knowledge base is proposed, aiming at achieving high compression ratio and high quality image reconstruction, and effectively solving the bandwidth limitation. This scheme segments the input image into several semantic information unit under the guidance of task requirements by Yolo-World and Segment Anything Model. The assigned bandwidth for each unit is according to the task relevance scores, which enables high-quality transmission of task-related information with lower communication overheads. An improved metric weighted learned perceptual image patch similarity (LPIPS) is proposed to evaluate the transmission accuracy of the novel scheme. Experimental results show that our scheme achieves a notable performance improvement on weighted LPIPS while the same compression ratio compared with traditional image compression schemes. Our scheme has a higher target capture ratio than traditional image compression schemes under the task of target detection.

An Unsupervised Image Fusion Approach Using Convolution Neural Network for Feature Enhancement of Medical Brain Images

Image fusion is the process of amalgamation of two or more images into a single image that contains composite, enriched with diverse details from the original images. In the medical field, image fusion serves as an indispensable tool for elevating the precision of medical imagery and facilitating diagnostic processes. With the advent of deep learning, there has been a significant increase in the accuracy and effectiveness of image fusion techniques. This paper presents a deeplearningbased approach for medical image fusion that combines the advantages of deep-learning techniques with traditional image fusion methods. The proposed method is evaluated on medical data from different modalities, and the experimental results show that the proposed method outperforms existing state-of-the-art image fusion techniques.

Feature fusion and context interaction for RGB-D indoor semantic segmentation

Depth information in indoor scenes is crucial to understanding the spatial and occluding relationships of objects. Depth information contains rich geometric details and is unaffected by lighting, texture, and color, significantly enhancing traditional color images. RGB images and depth images have been widely applied in various image analysis. However, effectively leveraging the complementary information from both modalities remains a challenge. To address this issue, we propose a unified and effective feature fusion and context interaction network (FCINet). It includes three modules. Firstly, we construct a multi-modal feature fusion module (MFFM), which can achieve the aggregation of two modalities along spatial and channel dimensions. Secondly, we construct a global and local information interaction context module (GLCM) to encode rich semantic information. Finally, we construct a feature alignment and fusion module (FAFM), which integrates upsampled low-level features with high-level features to mitigate spatial information loss. Experimental results indicate that the model achieved favorable outcomes on both the NYU Depth v2 dataset and the more complex SUN RGB-D dataset.

Efficient Metal Corrosion Area Detection Model Combining Convolution and Transformer

In the context of rapid industrialization, efficiently detecting metal corrosion areas has become a critical task in preventing material damage. Unlike conventional semantic segmentation targets, metal corrosion characteristics vary significantly in color, texture, and size. Traditional image segmentation methods need improvement in scenarios involving occlusions, shadows, and defects. This paper proposes a convolution and sequence encoding combined network, MCD-Net, for metal corrosion area segmentation. First, a visual Transformer sequence encoder is introduced into the convolutional encoder–decoder network to enhance global information processing capabilities and establish long-range feature dependencies. A feature fusion method based on an attention module is proposed to enhance the model’s ability to recognize corrosion boundaries, thereby enhancing segmentation accuracy and model robustness. Finally, in the model’s decoding stage, a score-based multi-scale feature enhancement method is employed to emphasize significant features in the corrosion areas. Experimental results indicate that this method attained an F1 score of 84.53% on a public corrosion dataset, demonstrating the model’s deeper understanding and reasoning capabilities for shadow and defect features, as well as excellent noise resistance performance.

Agricultural spray drone deposition, Part 1: Methods for high throughput spray pattern analysis

Abstract Effective pesticide application is dependent on precise and sufficient delivery of active ingredients to targeted pests. Water-sensitive papers (WSP) have been used to estimate the stain coverage, droplet density, droplet size, total spray volume, and other spray quality metrics by analyzing deposit stains using image analysis software. However, due to their expensive cost, WSPs are typically distributed along unidimensional transects at intervals of 0.5 m or more, which comprises 0.5% or less of the total treated area. This might limit the ability to accurately represent the deposition of agricultural sprayers with irregular patterns, such as agricultural drone sprayers in their early development stage. This study introduces a novel approach utilizing white Kraft paper and a blue colorant proxy for assessing spray deposition. A custom Python-based image analysis tool, SprayDAT (Spray Droplet Analysis Tool), was developed and compared to traditional image analysis software, DepositScan. Both models showed increased accuracy in detecting larger objects, with SprayDAT generally performing better for smaller droplets. DepositScan underestimated the total deposited spray volume by up to 2.7 times less compared to the colorant extraction assessed via spectrophotometry and the predicted output based on flow rate, coverage, and speed. Accuracy of software-estimated spray volume declined with increasing total stain coverage, likely due to overlapping stain objects. Droplet density exhibited a Gaussian trend with peak density at approximately 22% stain cover, offering evidence for overlapped stains for both DepositScan and SprayDAT as stain cover increased. Both models showed exponential growth in volumetric median diameter (VMD) with increasing stain cover. SprayDAT is freely accessible through an online repository. It features a user-friendly interface for batch-processing large sets of scanned images and offers versatility for customization to meet individual needs, such as adjusting spread factor, updating the standard curve for spray volume estimation, or modifying the stain detection threshold.

Non-invasive assessment of left ventricular hemodynamic forces in hypertrophic cardiomyopathy

Abstract Background Hemodynamic force (HDF) analysis allows non-invasive measurement of intraventricular pressure gradients, portraying cyclic spatial-temporal interaction between blood and tissues. Impaired HDFs have been implicated in early detecting of adverse cardiac remodelling and treatment response in various cardiovascular disorders, providing insights into cardiac physiology not offered by traditional cardiovascular imaging (1). Mainly explored in heart failure, HDF analysis has not been previously applied to hypertrophic cardiomyopathy (HCM). Purpose To investigate differences in systo-diastolic function in HCM patients compared to healthy controls using HDF analysis. Methods Forty patients with HCM diagnosis (20 obstructive and 20 non-obstructive) were retrospectively evaluated in comparison with 22 healthy controls. Left ventricular (LV) HDFs were derived from routine transthoracic apical 4-, 2- and 3-chamber views using a dedicated software. Apical-basal HDF curves were generated, where positive deflections represent forces directed from the LV apex to the base, whilst negative ones are directed toward the apex. Amplitude and timing parameters were derived, the latter indexed to total cardiac cycle duration. Results Table 1 shows the demographic, clinical and echocardiographic characteristics of HCM compared to controls. Patients with HCM were older, had a slower heart rate due to treatment with beta-blockers, and showed hypercontractility, as expressed by higher LV ejection fraction. Compared to controls, HCM patients showed reduced LV longitudinal force, i.e. the force during the whole cardiac cycle (4.17% [2.36-5.52] vs. 6.01% [4.73-7.78], p=0.002), shorter time interval to systolic peak (0.15 [0.12-0.16] vs. 0.19 [0.17-0.20], p <0.001), and shorter duration of LV impulse (0.29 [0.27-0.32] vs. 0.34 [0.32-0.37], p <0.001) (Figure 1). However, no differences in terms of systolic force peak (p=0.283) or LV impulse intensity (p=0.689) were detected. During the transition between systole and diastole, LV suction (i.e. the interval including late systolic deceleration and early diastolic suction) was significantly longer (0.26 [0.22-0.29] vs. 0.23 [0.21-0.25], p=0.023) but less pronounced (6.30% [4.56-8.44] vs. 8.72% [6.71-12.59], p=0.005), likely reflecting a relaxation impairment from the very beginning of diastole. Such impairment was maintained during early diastolic filling, where the amplitude of the positive deceleration flow force was severely reduced in HCM compared to controls (2.98% [1.98-4.58] vs. 7.11% [4.55-8.72], p <0.001). Conclusions HDF analysis in HCM patients revealed unique systolic and diastolic changes reflecting faster LV contraction during systole and slower and weaker LV forces during diastole. This analysis deepens our understanding of HCM mechanic abnormalities and may help assess response to innovative treatment options.

Correlation Hyperspectral Imaging

Hyperspectral imaging aims at providing information on both the spatial and the spectral distribution of light, with high resolution. However, state-of-the-art protocols are characterized by an intrinsic trade-off imposing to sacrifice either resolution or image acquisition speed. We address this limitation by exploiting light intensity correlations, which are shown to enable overcoming the typical downsides of traditional hyperspectral imaging techniques, both scanning and snapshot. The proposed approach also opens possibilities that are not otherwise achievable, such as sharper imaging and natural filtering of broadband spectral components that would otherwise hide the spectrum of interest. The enabled combination of high spatial and spectral resolution, high speed, and insensitivity to undesired spectral features shall lead to a paradigm change in hyperspectral imaging devices and open up new application scenarios. Published by the American Physical Society 2024

Fabric Defect Detection Method Based on Multi-scale Fusion Attention Mechanisms

Abstract Fabric defect detection is extremely important for the development of the textile industry, but the existing traditional image processing algorithms are not good enough to detect fabric defects, and the detection efficiency and accuracy of the classical deep learning model is not satisfactory, so this paper proposes an improved fabric defect detection method based on multi-scale fusion of attention mechanism YOLOv7-PCBS. Based on the YOLOv7 network structure, some of the standard convolutions of the backbone network are replaced with Partial Convolution (PConv) modules, which reduces the amount of network computation and improves the network detection speed; add Coordinate Attention (CA) to enhance the ability of extracting the positional features of tiny defects in fabrics; reconfiguration of the SPPCSPC module to improve small target detection; optimization of Bidirectional Feature Pyramid Network (BiFPN) and design of Tiny- BiFPN for simple and fast multi-scale feature fusion; finally, a novel loss function SIoU with angular loss is introduced to facilitate the fitting of the true and predicted frames and enhance the accuracy of defect prediction. The results show that the algorithm achieves a mAP value of 94.4% on the detection of defects in solid-colored fabrics of six denim materials, which is an improvement of 15.1% compared to the original YOLOv7 algorithm, while the model achieves a frame rate of 59.5 per second. Compared with other traditional deep learning algorithms SSD and Faster-RCNN, the detection accuracies are improved by 21.6% and 15.2%, and the FPS values are improved by 78.1% and 101.0%, respectively. Therefore, the YOLOv7-PCBS fabric defect detection algorithm proposed in this paper makes the fabric defect detection results more accurate while realizing lightweight, which provides an important technical reference for the subsequent improvement of textile quality.

YOLO-DroneMS: Multi-Scale Object Detection Network for Unmanned Aerial Vehicle (UAV) Images

In recent years, research on Unmanned Aerial Vehicles (UAVs) has developed rapidly. Compared to traditional remote-sensing images, UAV images exhibit complex backgrounds, high resolution, and large differences in object scales. Therefore, UAV object detection is an essential yet challenging task. This paper proposes a multi-scale object detection network, namely YOLO-DroneMS (You Only Look Once for Drone Multi-Scale Object), for UAV images. Targeting the pivotal connection between the backbone and neck, the Large Separable Kernel Attention (LSKA) mechanism is adopted with the Spatial Pyramid Pooling Factor (SPPF), where weighted processing of multi-scale feature maps is performed to focus more on features. And Attentional Scale Sequence Fusion DySample (ASF-DySample) is introduced to perform attention scale sequence fusion and dynamic upsampling to conserve resources. Then, the faster cross-stage partial network bottleneck with two convolutions (named C2f) in the backbone is optimized using the Inverted Residual Mobile Block and Dilated Reparam Block (iRMB-DRB), which balances the advantages of dynamic global modeling and static local information fusion. This optimization effectively increases the model’s receptive field, enhancing its capability for downstream tasks. By replacing the original CIoU with WIoUv3, the model prioritizes anchoring boxes of superior quality, dynamically adjusting weights to enhance detection performance for small objects. Experimental findings on the VisDrone2019 dataset demonstrate that at an Intersection over Union (IoU) of 0.5, YOLO-DroneMS achieves a 3.6% increase in mAP@50 compared to the YOLOv8n model. Moreover, YOLO-DroneMS exhibits improved detection speed, increasing the number of frames per second (FPS) from 78.7 to 83.3. The enhanced model supports diverse target scales and achieves high recognition rates, making it well-suited for drone-based object detection tasks, particularly in scenarios involving multiple object clusters.