Image Processing Vision Research Articles

The Ground-Penetrating Radar Image Matching Method Based on Central Dense Structure Context Features

Subsurface structural distribution can be detected using Ground-Penetrating Radar (GPR). The distribution can be considered as road fingerprints for vehicle positioning. Similar to the principle of visual image matching for localization, the position coordinates of the vehicle can be calculated by matching real-time GPR images with pre-constructed reference GPR images. However, GPR images, due to their low resolution, cannot extract well-defined geometric features such as corners and lines. Thus, traditional visual image processing algorithms perform inadequately when applied to GPR image matching. To address this issue, this paper innovatively proposes a GPR image matching and localization method based on a novel feature descriptor, termed as central dense structure context (CDSC) features. The algorithm utilizes the strip-like elements in GPR images to improve the accuracy of GPR image matching. First, a CDSC feature descriptor is designed. By applying threshold segmentation and extremum point extraction to the GPR image, stratified strip-like elements and pseudo-corner points are obtained. The pseudo-corner points are treated as the centers, and the surrounding strip-like elements are described in context to form the GPR feature descriptors. Then, based on the feature description method, feature descriptors for both the real-time image and the reference image are calculated separately. By searching for the nearest matching point pairs and removing erroneous pairs, GPR image matching and localization are achieved. The proposed algorithm was evaluated on datasets collected from urban roads and railway tracks, achieving localization errors of 0.06 m (RMSE) and 1.22 m (RMSE), respectively. Compared to the traditional Speeded Up Robust Features (SURF) visual image matching algorithm, localization errors were reduced by 86.6% and 95.7% in urban road and railway track scenarios, respectively.

Research on motion control strategy of athlete muscle training based on blockchain and visual image analysis

Muscle training is an important part of athletes’ physical training. Its goal is to keep athletes in high intensity and improve sports performance in physical training. With the continuous improvement of the level of competitive sports, there are still many problems in the training of competitive athletes. The research on the control strategies of some technical movements in the process of athletes’ muscle development would help to improve the efficiency and effect of athletes’ training. By analyzing and quant description of muscle movement control process based on blockchain and visual image processing technology, the muscle movement control rule is discussed and then a complete movement database is constructed. At the same time, combined with the sports experiment method and cognitive behavioral theory, this paper analyzed the exercise load, exercise time allocation, muscle training time allocation methods and the corresponding muscle training strategy selection methods under different muscle states, providing theoretical basis for athletes to conduct muscle strength training scientifically and efficiently. This paper first analyzed the importance of athletes’ muscle training action control. After that, the research was mainly carried out from the following two aspects. The first is to analyze the motion control strategy of athlete muscle training based on blockchain and visual image, and summarize its characteristics. The second is to use electronic imaging technology to analyze the data related to posture control in athletes’ muscle training, and explore the characteristics and changes of muscle posture. At last, this paper put forward relevant algorithms about visual images and conducted experimental research on athletes’ physical indicators and performance under different muscle training according to the research in this paper. It was concluded that the score ratio of athletes’ performance after muscle training action control was 7.39% higher than that of general muscle training. Therefore, it is very important to strengthen the research of movement technology in muscle training and analyze the influence of some training control strategies on the development of athletes’ muscles in training.

Research on the Knowledge Structure and Sustainable Development Pathways of Artificial Intelligence from the Perspective of Technological Science

Achieving significant breakthroughs in both the fundamental theories and technological applications of artificial intelligence is essential for fostering its long-term development. Under the guidance of Professor Qian Xuesen’s theory of technological science, exploring the internal mechanisms of knowledge evolution in artificial intelligence holds profound theoretical and practical significance for promoting sustainable technological advancement. This study draws on literature from the Web of Science (WOS) database and employs methods such as knowledge mapping, natural language processing, clustering analysis, and citation analysis to outline the knowledge structure of the field, clarify the trajectory of sustainable development, and trace the technological genealogy of VR/AR technologies.This study divides the knowledge structure within the field of technological science into “basic theoretical knowledge—applied basic knowledge—applied knowledge”, enriching Qian’s theory of technological science from within and providing strong intellectual support and technological pathways for sustainable technological development in practice. Artificial intelligence encompasses 10 distinct knowledge domains, among which machine learning and deep learning constitute the basic theoretical knowledge, data intelligence, computer vision, and swarm intelligence are the applied basic knowledge, and image processing and human-computer intelligence are the applied knowledge. The development of VR/AR technology has formed two main sustainable development paths: “machine learning—data intelligence—intelligent systems—human computer intelligence”, and “deep learning—computer vision—image processing”.

Coupled thermal-electrical–mechanical characteristics of lightning damage in woven composite honeycomb sandwich structures

In this study, lightning strike damage of woven carbon fibre-reinforced polymer laminates (W-CFRPs) and woven composite honeycomb sandwich panels (W-CHSPs) are simulated using the proposed sequential thermal-electrical–mechanical finite element (FE) coupling model incorporating dielectric breakdown of materials. Surface current with an amplitude of 200 kA and corresponding lightning shockwave overpressure were applied on each composite. The FE model coupled with LaRC05 criterion was used to study the failure behaviours of intralaminar damage and interlaminar delamination of the W-CFRPs and W-CHSPs. A series of lightning strike tests were performed to validate the FE model. Detailed lightning damage assessments and mechanisms were characterized by a combination of visual inspection, image processing, ultrasonic scanning and micro computed tomography (Micro-CT) and showed good agreements with the FE-predicted results. It can be concluded that shockwave overpressure significantly impacts lightning-induced damages, thereby supporting the effectiveness of the newly proposed sequential thermal-electrical–mechanical coupling model, which demonstrates improved predictive accuracy.

Multi-disease X-ray Image Classification of the Chest Based on Global and Local Fusion Adaptive Networks

Chest X-ray image classification for multiple diseases is an important research direction in the field of computer vision and medical image processing. It aims to utilize advanced image processing techniques and deep learning algorithms to automatically analyze and identify X-ray images, determining whether specific pathologies or structural abnormalities exist in the images. We present the MMPDenseNet network designed specifically for chest multi-label disease classification. Initially, the network employs the adaptive activation function Meta-ACON to enhance feature representation. Subsequently, the network incorporates a multi-head self-attention mechanism, merging the conventional convolutional neural network with the Transformer, thereby bolstering the ability to extract both local and global features. Ultimately, the network integrates a pyramid squeeze attention module to capture spatial information and enrich the feature space. The concluding experiment yielded an average AUC of 0.898, marking an average accuracy improvement of 0.6% over the baseline model. When compared with the original network, the experimental results highlight that MMPDenseNet considerably elevates the classification accuracy of various chest diseases. It can be concluded that the network, thus, holds substantial value for clinical applications.

UAV forest fire detection based on CNN and InceptionV3

Abstract Forest fires have created a severe hazard in our ecological environment with abrasive climate change all over the world. What is even more concerning, the frequency of forest fires is increasing year by year. By far, a great number of methodologies, such as satellite systems, IR, and WSN, have been created to resolve this rigorous problem. However, most of these detection methods are costly and structurally complex, making them difficult to quickly learn and use, thus not effectively reducing personnel and property injuries. Our research proposed using UAVs (Unmanned Aerial Vehicles) as visual detection carriers and using InceptionV3 for visual processing of aerial images. The flame dataset from IEEE Dataport, containing 63178 JPEG pictures, is used in our research. Our proposed forest fire detection methodology using Convolutional Neural Network (CNN) and InceptionV3 achieved a remarkable 89.90% validation accuracy. Moreover, our methodology gained an ideal precision and recall ratio.

Classification of wheat flour levels in powdered spices using visual imaging

The present research aims to investigate the ability of image processing in the detection of wheat flour adulterant in different spices. Different samples of black pepper, red pepper, and cinnamon were prepared with different levels of wheat flour adulteration. The images of the samples were captured using a visible camera. An image processing algorithm was designed and codded in MATLAB software to process the acquired images. Different color and textural features were extracted. The efficient features were selected and classified using artificial neural network and support vector machine. In the present research, the classification accuracies of artificial neural network analysis for black pepper, red pepper, and cinnamon products were 97.8, 100.00, and 100.00 %, respectively. During the experiment, these results were better than the performance of the support vector machine, (93.33, 100.00, and 98.88 %, respectively). Visible image processing can be used accurately to detect different adulterant levels that increase product quality while reducing operating costs.

Hetero-associative Memory Based New Iraqi License Plate Recognition

نتيجة للتطورات الأخيرة في أبحاث الطرق السريعة بالإضافة إلى زيادة استخدام المركبات، كان هناك اهتمام كبير بنظام النقل الذكي الأكثر حداثة وفعالية ودقة (ITS) في مجال رؤية الكمبيوتر أو معالجة الصور الرقمية، يلعب تحديد كائنات معينة في صورة دورًا مهمًا في إنشاء صورة شاملة. هناك تحدٍ مرتبط بالتعرف على لوحة ترخيص السيارة (VLPR) بسبب الاختلاف في وجهة النظر، والتنسيقات المتعددة، وظروف الإضاءة غير الموحدة في وقت الحصول على الصورة والشكل واللون، بالإضافة إلى الصعوبات مثل ضعف دقة الصورة ، الصورة الباهتة ، الإضاءة السيئة، التباين المنخفض، يجب التغلب عليها. اقترحت هذه الورقة نموذجًا باستخدام تعديل الذاكرة الترابطية ثنائية الاتجاه (MBAM)، وهي نوع واحد من الذاكرة الترابطية غير المتجانسة، وتعمل MBAM على مرحلتين)مرحلتي التعلم والتقارب) للتعرف على اللوحة، ويمكن لهذا النموذج المقترح التغلب على تلك الصعوبات بسبب قدرة الذاكرة الترابطية لـ MBAM على قبول الضوضاء وتمييز الصور المشوهة، وكذلك سرعة عملية الحساب نظرًا لصغر حجم الشبكة. نتيجة دقة تحديد منطقة اللوحة هي 99.6٪، ودقة تجزئة الأحرف 98٪، والدقة المحققة للتعرف على الأحرف هي100 ٪ في ظروف مختلفة.

Detector-embedded reconstruction of complex primitives using FPGAs

The slowdown of Moore’s law and the growing requirements of future HEP experiments with ever-increasing data rates pose important computational challenges for data reconstruction and trigger systems, encouraging the exploration of new computing methodologies. In this work we discuss a FPGA-based tracking system, relying on a massively parallel pattern recognition approach, inspired by the processing of visual images by the natural brain (“retina architecture”). This method allows a large efficiency of utilisation of the hardware, low power consumption and very low latencies. Based on this approach, a device has been designed within the LHCb Upgrade-II project, with the goal of performing track reconstruction in the forward acceptance region in real-time during the upcoming Run 4 of the LHC. This innovative device will perform track reconstruction before the event-building, in a short enough time to provide pre-reconstructed tracks (“primitives”) transparently to the processor farm, as if they had been generated directly by the detector. This allows significant savings in higher-level computing resources, enabling handling higher luminosities than otherwise possible. The feasibility of the project is backed up by the results of tests performed on a realistic hardware prototype, that has been opportunistically processing actual LHCb data in parallel with the regular DAQ in the LHC Run 3.

Energy-efficient real-time visual image adversarial generation and processing algorithm for new energy vehicles

Energy-efficient real-time visual image adversarial generation and processing algorithm for new energy vehicles

Perception-Driven Soft-Edge Occlusion for Optical See-Through Head-Mounted Displays.

Systems with occlusion capabilities, such as those used in vision augmentation, image processing, and optical see-through head-mounted display (OST-HMD), have gained popularity. Achieving precise (hard-edge) occlusion in these systems is challenging, often requiring complex optical designs and bulky volumes. On the other hand, utilizing a single transparent liquid crystal display (LCD) is a simple approach to create occlusion masks. However, the generated mask will appear defocused (soft-edge) resulting in insufficient blocking or occlusion leakage. In our work, we delve into the perception of soft-edge occlusion by the human visual system and present a preference-based optimal expansion method that minimizes perceived occlusion leakage. In a user study involving 20 participants, we made a noteworthy observation that the human eye perceives a sharper edge blur of the occlusion mask when individuals see through it and gaze at a far distance, in contrast to the camera system's observation. Moreover, our study revealed significant individual differences in the perception of soft-edge masks in human vision when focusing. These differences may lead to varying degrees of demand for mask size among individuals. Our evaluation demonstrates that our method successfully accounts for individual differences and achieves optimal masking effects at arbitrary distances and pupil sizes.

A Case Study of Visual Detection and Image Processing Courses under the OBE Teaching

A Case Study of Visual Detection and Image Processing Courses under the OBE Teaching

GSANet: Enhancing accuracy in the segmentation of challenging medical images

Medical image segmentation is crucial in the fields of computer vision and medical image processing. Its difficulty lies in precisely extracting regions of interest from images, such as blood vessels, tumors, or other anatomical structures, which is especially important in retinal images and skin pathology images. However, factors like noise, low contrast, and the complexity of target structures in medical images severely affect the accuracy of segmentation.Despite significant progress made by deep learning models such as U-Net in medical image segmentation, their performance still needs improvement when dealing with medical images with complex contexts and long-range dependencies. To tackle these challenges, we introduce a new deep learning model named GSANet.The GSANet model retains the advantages of U-Net while incorporating two innovative modules to enhance its performance in handling challenging medical image segmentation tasks. We assess the GSANet model's performance on two demanding medical image segmentation tasks, DRIVE and ISIC2018. Experimental findings illustrate that the GSANet model achieves significant performance enhancements on these tasks, surpassing the current state-of-the-art methods.The success of the GSANet model can be attributed to its effective design and implementation. GSANet excels in handling medical image segmentation tasks with complex contexts and long-range dependencies, leading to superior performance. Our research highlights the superiority of the GSANet model in medical image segmentation tasks, opening up new possibilities for further improvements in the performance and accuracy of medical image segmentation.

Analysis of stability and internal flow mechanisms in spiral annular flow with different swirlers

Spiral annular flow within ducts is widely utilized in modern industry, with the swirler serving as a critical component in generating such flow patterns. The structure of the swirler significantly influences the generation and stability of the spiral annular flow. This study selected four different swirler structures with outstanding performance from previous research and analyzed their characteristics through visual image processing combined with numerical simulations. By analyzing the amplitude information of the wave fluctuations in the annular swirling flow liquid film under different operating conditions using probability density functions, it was found that the swirler A(Flat-vane swirler) and swirler B(Flat-vane swirler with hub) produced smaller fluctuations in the annular swirling flow liquid film, indicating better stability compared to the swirler C(Arc-vane swirler) and swirler D(Spiral-vane swirler), which exhibited poor performance. Combining the numerical simulation results with the analysis of the internal mechanism of the swirlers, it was discovered that within the swirler A and swirler B, the fluid between the swirler vanes experienced a larger pressure gradient, resulting in phenomena such as “jump” and “pull” under this pressure gradient. This, in turn, contributed to the generation of greater tangential velocity and radial pressure gradient after the fluid exited the swirler. Due to the influence of the swirler structure, the swirler A and swirler B did not completely separate the fluid region into four independent spaces. Instead, in the central connection area of the rear section of the swirler, the gas phase components aggregated earlier, greatly promoting the downstream generation of spiral annular flow. This study analyzed the two-phase flow process and mechanism inside the swirler, filling a gap in previous research and providing important references for the optimization and selection of swirlers.

Deep Learning for Computer Vision: Innovations in Image Recognition and Processing Techniques

Deep learning is a key area of research in the field of computer vision, image processing and bioinformatics. The techniques of deep learning generally are divided into three categories namely Convolutional Neural Networks (CNN), Restricted Boltzmann Machines (RBM), Stacked RBM and HOG (Histograms of oriented Gradient) feature extraction, Convolutional Neural Networks as a Database (CNN as D). Additionally, one in few deep learning architectures which is gaining popularity and is frequently used in the field of computer vision and image processing is Extreme Learning Machine and ensemble of Extreme Learning Machine and CNN. It attempts to survey the recent advances in deep learning researchers and the application of these algorithm in the field of computer vision. Mainly focusing on the deep learning methods and algorithms rather than image processing and computer vision methods, this work inspects deep learning techniques which are widely and commonly used in the field of computer vision image detection and processing like CNN, DBN, RBM and HMM as well as various applications of these techniques. Applications of deep learning techniques in computer vision are image classification, object recognition and detection. Along with the recent works and the future scope for deep learning methods in the field of computer vision and image processing is presented.

Integrate Large Language Models (LLMs) with Computer Vision in the Field of Robotics for Computer Vision Image Processing, Simultaneous Localization and Mapping and to Prevent/Secure Robot Malfunctions

The development of robotics has affected many parts of the world, but the application of robotic systems in complex environments is still a concern. This paper proposes GANs and robotics computer vision to enhance image processing and SLAM, which are key for autonomous navigation. The methodology quantitatively compares different SLAM systems using various environments, including traditional ones and those supplemented by GANs. There is a higher accuracy of localization and mapping, better performance in complex environments with abundant visual elements, and higher computational efficiency. This paper will present an innovative application of GANs in robotics vision and, in the process, establish a new benchmark for how robotic systems perceive the actual world. This study, therefore, is part of the advancement of robotic systems. It provides a novel approach to using GANs as a robot vision tool and offers empirical evidence that the traditional robotic platform can be integrated with machine learning.

Enhancing computer image recognition with improved image algorithms

Advances in computer image recognition have significantly impacted many industries, including healthcare, security and autonomous systems. This paper aims to explore the potential of improving image algorithms to enhance computer image recognition. Specifically, we will focus on regression methods as a means to improve the accuracy and efficiency of identifying images. In this study, we will analyze various regression techniques and their applications in computer image recognition, as well as the resulting performance improvements through detailed examples and data analysis. This paper deals with the problems related to visual image processing in outdoor unstructured environment. Finally, the heterogeneous patterns are converted into the same pattern, and the heterogeneous patterns are extracted from the fusion features of data modes. The simulation results show that the perception ability and recognition ability of outdoor image recognition in complex environment are improved.

STUDY OF IMAGE SEGMENTATION METHODS WITH MRI IMAGES

Digital image processing is the use of a digital computer to process digital images. Image processing transforms input images into digital form for certain operations to obtain useful information. Segmentation is a well-known process used in image processing that partitions input images into different regions. Image segmentation is a sub-area of computer vision and digital image processing for grouping similar segments of an image under respective class labels. Several methods were performed with neutrosophic sets on dissimilar image-processing domains. However, the denoising and segmentation were not carried out accurately with minimal time complexity. To address these issues, many image segmentation methods are reviewed.

Application of machine vision image processing technology in dental implant surgery

Machine vision image processing technology is extensively employed in the medical realm, particularly in dynamic navigation and robotic systems for oral implantology. It plays a pivotal role in assisting clinicians with precise implant placements, enhancing the predictability of implant restorations. The fundamental principles of machine vision image processing technology utilized in dynamic navigation and robotic systems for oral implantology primarily encompass spatial positioning and registration. However, due to variations in technical principles among different systems, their workflows and technical nuances exhibit distinctive characteristics. Therefore, commencing from the principles of spatial positioning and registration in machine vision image processing technology. This article delves into the current application landscape of machine vision in dynamic navigation and robotics for oral implantology. Its objective is to furnish valuable insights for the clinical implementation of machine vision-assisted implant technology.

Demonstration of computer vision for void characterisation of 3D-printed continuous carbon fibre composites

Microstructural void formation during the processing of continuous fibre-reinforced polymer composites are a significant limitation of processes such as 3D printing, as voids inhibit resultant mechanical properties. Traditional optical microscopy approaches to void characterisation are tedious and are prone to errors due to non-ideal but realistic image conditions, including low contrast typical of microscopic images. This paper proposes a novel application of automated computer vision image processing techniques (i.e. a contrast-limited adaptive histogram equalisation (CLAHE) algorithm and Otsu's method) for detailed void characterisation of continuous fibre-reinforced composites. Microstructural void fraction was determined for both unidirectional and bi-axial laminates of a variety of thicknesses. The void characterisation by computer vision was shown to produce comparable results with manual contrast/brightness control, but with significantly less standard deviation. The computer vision software used in void characterisation is made freely available online at https://github.com/Xiao-ying/VoidDetector.