Machine Vision Algorithms Research Articles

From photographic images to hierarchical networks―Color associations of a traditional Chinese garden

AbstractThe complexity of the environment and the richness of landscape color make it difficult to objectively measure color in gardens. However, with the use of machine vision and image processing algorithms, it is now possible to accurately quantify the colors of the built environment through photo recognition, classification, enhancement, and segmentation. This approach can produce correlation intensity charts and hierarchical networks that illustrate theme colors and their associations. To explore this methodology, we have selected Beijing Xiangshan Temple―a traditional Chinese garden and temple with rich landscape colors―as the research object. Using a combination of machine vision image processing technology and color clustering algorithms, we aim to establish a landscape color analysis model based on the Munsell color system and color harmony theory. Ultimately, this will provide a quantifiable and objective reference basis for the design and repair of temple garden colors, with a view to achieving the inheritance and development of classical gardens.

Design and Implementation of a Dynamic Tracking System for Drones in Modern Medical Treatment Applications

The utilization of unmanned aerial vehicles (UAVs) equipped with advanced machine vision and remote sensing technology has ushered in a transformative era across diverse domains such as agriculture, construction, energy, environmental monitoring, and contemporary medical treatment. Particularly within the realm of modern medical treatment, the integration of drone technology into healthcare service delivery has displayed tremendous promise for enhancing patient outcomes, particularly in geographically challenging areas or during emergent situations. Quad-rotor UAVs have emerged as prominent contenders owing to their exceptional performance, user-friendly controls, and adaptable functionality. Their quad-rotor configuration enables seamless vertical takeoff and landing, rendering them highly suitable for deployment in demanding environments frequently encountered in the context of modern medical treatment. When combined with cutting-edge machine vision and remote sensing capabilities, quad-rotors exhibit proficiency in executing intricate missions encompassing medical supply delivery, mobile target tracking, anomaly detection, topographical mapping, and even facilitating emergency medical response scenarios. These remarkable capabilities render quad-rotors indispensable tools for healthcare professionals including doctors, nurses, and other dedicated personnel in the provision of high-quality healthcare services within the realm of modern medical treatment. The design and implementation of a dynamic tracking system predicated on machine vision and remote sensing technology assume paramount importance in ensuring optimal UAV operations in modern medical treatment applications. A meticulously crafted system should possess the capacity to process visual information through sophisticated machine vision algorithms, thereby transforming it into precise control data for the UAV. Employing a remote sensing target detection model serves to accurately locate and identify pertinent information embedded within optical remote sensing imagery, thereby generating a comprehensive output feature map information set. Integration of this information with the cascade PID (Proportion Integral Differential) control algorithm facilitates the realization of a dynamic tracking system tailored specifically for quad-rotor UAVs. Through meticulous experimentation, the system can be effectively harnessed for portable high-altitude visual detection, thereby endowing healthcare services with promptness and efficiency. For instance, the system can be leveraged to meticulously track the movement of medical supplies during transportation, remotely monitor patients’ vital signs, and even facilitate the expeditious delivery of essential medications or vaccines to individuals situated in remote and inaccessible locations during modern medical treatment. As the demand for healthcare services continues to surge, the prevalence of drones equipped with state-of-the-art machine vision and remote sensing technology within modern medical treatment contexts is expected to increase manifold. Consequently, sustained research and development endeavors dedicated to advancing this technology assume paramount significance, thereby ensuring its seamless and efficacious implementation within the modern medical treatment field.

Research and Development of Machine Vision Algorithm Performance Evaluation System in Complex Scenes

In recent years, deep learning technology has gained widespread attention and has been successfully applied in various applications, including machine vision for automatic driving. However, one of the major challenges in this field is the ability of AI models to perform well in complex scenes with varying weather conditions and lighting conditions. The proposed scheme uses VR technology to simulate different weather conditions and lighting conditions and to evaluate the performance of AI products in such complex scenes. To further enhance the generalization capabilities of AI models, the Software Development Kit has been updated with complex scene data with configurable parameters, and training data with typical characteristics according to machine vision tasks. This approach has the potential to significantly improve the accuracy and reliability of machine vision systems for automatic driving, ultimately enhancing their safety and effectiveness.

Machine Vision Algorithm for Identifying Packaging Components of HN-3 Arterial Blood Sample Collector

The arterial blood sample collector produced in large quantities often fails to meet the requirements due to missing components in the packaging bag, and traditional manual detection methods are both inefficient and inaccurate. To solve this problem, a PyCharm-integrated development environment was used to study image processing and recognition algorithms for identifying components inside the packaging bag of the HN-3 arterial blood sample collector. The machine vision system was used to capture images of the packaging bags of the HN-3 Arterial blood sample collector. Template matching was employed to extract the packaging ROI, and the threshold segmentation method in the HSV color model was used to extract material features based on the packaging ROI. Morphological processing algorithms such as dilation or erosion were used to enhance the connectivity of the extracted features. The existence of components was determined by setting thresholds for the connected domain area or length. The results of the recognition experiment show that the false detection rate is 0.2%, the missed detection rate is 0%, and the average image processing time per product is no more than 39 ms. Compared with manual recognition methods, the efficiency and accuracy have been improved by 36.5 times and 2.3%, respectively. The experimental results confirm the effectiveness of the image processing algorithm.

Design of mountain fire prevention monitoring system for transmission lines based on machine vision algorithms

Abstract With the continuous development of society, power grid has become a major guarantee for social and economic development, and its coverage is also expanding. The power grid not only covers some developed areas, but also extends to remote mountainous areas with complex changes in geological and climatic conditions. However, at the same time, line failures and power outages caused by mountain fires are becoming increasingly frequent, which poses a significant threat to the safety and stable operation of the power system. Therefore, how to solve the problem of mountain fire on transmission lines has become an important topic today. The main purpose of this study is to achieve effective monitoring of transmission line mountain fires and provide more assurance for the stable operation of the power system. This article discussed the transmission line and traditional mountain fire prevention monitoring system, and combined the current hot machine vision algorithm to design the overall mountain fire prevention monitoring system, focusing on analyzing the differences between this system and traditional systems. Through experiments, it was found that the monitoring accuracy of the mountain fire prevention monitoring system designed in this article has been improved by 10.6 % compared to the traditional system at location 4, 7.6 % at location 6, and 7.9 % at location 7. At the same time, the real-time performance and anti-interference performance of this system were also better than traditional systems, and the overall performance of the system was better. Practice has proved that the system can effectively overcome the interference of external environment, and monitor and identify the fire situation in real time and accurately, ensuring the safety of transmission lines. At the same time, the research on machine vision algorithms in this paper can also help promote the more perfect development of power systems. Practice has proved that this system can effectively overcome the interference of the external environment, and continuously and accurately detect and identify the fire situation.

Research on defect detection method of bearing dust cover based on machine vision and multi-feature fusion algorithm

During the assembly process of deep groove ball bearings, due to defective parts and unqualified assembly process, various indentations and scratches on the dust cover will often result in reducing the service life and reliability of the bearing. Therefore, the online monitoring of the assembly quality of the dust cover ensures the necessary detection process of the bearing surface quality. This paper proposed a bearing dust cover defect detection method based on machine vision and multi-feature fusion algorithm, which can effectively detect bearings with dust cover defects. The algorithm first performs Laplace transform and Sobel operator image enhancement on the collected bearing images. Extract and fuse multi-source fault feature with the scale-invariant feature transform (SIFT), bag-of-visual-words (BoVW) and GLCM-Hu methods. Machine learning and deep learning models were constructed, and the performance of each model was compared through feature visualization and misclassified analysis. The results show that the extracted multi-source features are more representative and robust. The SIFT-BoVW-GS-SVM model achieved the best detection results in detecting bearing dust cover defects with an accuracy of 91.11%. The processing and program detection time for each bearing image is about 0.019 s. The accuracy and speed of detection and judgment meet the needs of online defect detection of bearing dust cover.

An automated optical inspection (AOI) platform for three-dimensional (3D) defects detection on glass micro-optical components (GMOC)

With the widespread deployment of wavelength division multiplexing (WDM), optical transceivers increasingly use many glass micro-optical components (GMOC). Visual inspection of these GMOCs is a critical manufacturing step to ensure quality and reliability. However, manual inspection is often labor-intensive and time-consuming due to the transparent nature of glass components and the small, randomly located defects in three dimensions. Although automated optical inspection (AOI) exists, it has not yet been able to provide the desired level of accuracy and efficiency.This paper reports the development of an AOI platform for 3D defect detection on GMOCs. The platform incorporates 3D video acquisition and a novel two-stage neural network machine-learning algorithm. It includes a robotic arm for moving parts in 3D, a camera with an illumination module for video acquisition, and a video streaming processing unit with a machine vision algorithm for real-time defect detection on a production line. The robotic arm enables multi-perspective video capture of a test sample without refocusing. The two-stage machine learning network uses a modified YOLOv4 architecture with color channel separation (CCS) convolution, an image quality evaluation (IQE) module, and a frame fusion module to integrate the single frame detection results. This network can process multi-perspective video streams in real-time for defects detection in a coarse-to-fine manner. The AOI platform was trained with only 30 samples and achieved promising performances with a recall rate of 1, a detection accuracy of 97%, and an inspection time of 48 s per part.

Enhancing human sight perceptions to optimize machine vision: Untangling object recognition using deep learning techniques

The goal of machine vision is to develop human-like visual abilities; however, it is unclear whether understanding human vision will advance machines. Here, it exemplifies two key conceptual advancements: It first shows that the majority of computer vision models consistently differ from the way that individuals perceive objects. To do this, a significant dataset of human perceptions of the separations of isolated things was acquired, and it was then examined to see if a well-known machine vision algorithm can predict these perceptions. The best algorithms can account for the majority of the volatility in the intuitive data, but every algorithm we verified repeatedly misjudged several different object types. Second, it shows that removing these systemic biases can considerably increase classification accuracy. For instance, machine techniques overestimated detachments between symmetric objects in comparison to human vision. These results illustration that methodical differences between human and machine vision can be identified and improved.In order to improve the machine vision, employing a deep learning algorithm Visual Geometry Group (VGG 16) with planar reflection symmetry (PRS-Net) technique. VGG 16 is a convolutional neural network with 16 deep layers. VGG pre-trained architecture can point out visual features present in the image. The planar reflection symmetry concept is appended with VGG to create a hybrid environment that can improve machine vision significantly by 90%.

A Machine-Vision Approach to Transmission Electron Microscopy Workflows, Results Analysis and Data Management.

Transmission electron microscopy (TEM) enables users to study materials at their fundamental, atomic scale. Complex experiments routinely generate thousands of images with numerous parameters that require time-consuming and complicated analysis. AXON synchronicity is a machine-vision synchronization (MVS) software solution designed to address the pain points inherent to TEM studies. Once installed on the microscope, it enables the continuous synchronization of images and metadata generated by the microscope, detector, and in situ systems during an experiment. This connectivity enables the application of machine-vision algorithms that apply a combination of spatial, beam, and digital corrections to center and track a region of interest within the field of view and provide immediate image stabilization. In addition to the substantial improvement in resolution afforded by such stabilization, metadata synchronization enables the application of computational and image analysis algorithms that calculate variables between images. This calculated metadata can be used to analyze trends or identify key areas of interest within a dataset, leading to new insights and the development of more sophisticated machine-vision capabilities in the future. One such module that builds on this calculated metadata is dose calibration and management. The dose module provides state-of-the-art calibration, tracking, and management of both the electron fluence (e-/Å2·s-1) and cumulative dose (e-/Å2) that is delivered to specific areas of the sample on a pixel-by-pixel basis. This enables a comprehensive overview of the interaction between the electron beam and the sample. Experiment analysis is streamlined through a dedicated analysis software in which datasets consisting of images and corresponding metadata are easily visualized, sorted, filtered, and exported. Combined, these tools facilitate efficient collaborations and experimental analysis, encourage data mining and enhance the microscopy experience.

Determination of the error of measurements obtained by the optical micrometry

We present the results of measuring the volume of polyvinyl alcohol polymer granules crosslinked with epichlorohydrin in water and in aqueous solutions of KCl, MgCl2 and their mixtures, obtained by optical micrometry, and consider the main sources of errors in the measurement errors. The purpose of this study is to analyze and evaluate the effect of the main sources of errors on the accuracy of determining the relative volumes of granules, as well as to search for techniques that can minimize the resulting measurement errors. The diameters of the granules were determined using specialized software implementing machine vision algorithms from the images obtained by optical microscopy. Their volumes were calculated using the formula for the volume of the ellipsoid of revolution. The maximum accuracy of volume determination is known to be achieved when the measured granule has a sphere shape. It is shown that deviation from this shape, for example, in case of an ellipsoid, gives errors in determining the third axis of the ellipsoid, invisible in the image, which creates an error in determining the relative volume of the granule. The instrument error is determined and a statistical estimate of the error attributed to the non-sphericity of the granules is given. It is shown that a typical instrument error in determining the relative volumes of granules is 0.4%. The non-sphericity of the measured granules increases the measurement error up to 3.5%. The error for a single granule can be reduced to 2.3% by combination of methodological techniques and statistical processing of the results, whereas and for an ensemble of at least 5 granules — up to 1.5%. The reproducibility of the properties of polymer granules in cyclic measurements was studied. It is shown that the degree of swelling the granules is reproduced with an error of 1%, which allows the sensor to be used repeatedly. The results obtained can be used in experiments and data processing for analytical applications.

QR code: The global making of an infrastructural gateway

This article traces the history of machine-readable data encoding standards and argues that the QR code has become an infrastructural gateway. Through the analysis of patents, corporate documents and advertising, ethnographic observations, and interviews with professionals, I describe the global making of the QR code and argue that the convergence of data encoding standards, mobile computing, machine vision algorithms, and platform ecosystems has led to the emergence of a new component of computational infrastructures which functions as a gateway between different actors, systems, and practices. The central section of the article covers seven decades of machine-readable data encoding history across different national and regional contexts: from the invention and popularization of the barcode in the United States, through the QR code’s invention in Japan and its success in East Asia, to its platformization in China. By revisiting this history through concepts drawn from the field of infrastructure studies, I argue that QR codes have become infrastructural gateways and conclude that this concept is useful not only to understand the current role of QR codes but also to identify and follow the emergence and change of other gateways in infrastructures to come.

Color Machine Vision Design Methodology of a Part-Presentation Algorithm for Automated Poultry Handling

This article offers an image processing method to optimize the design of a three-level color machine vision algorithm illustrated in the context of presenting a whole chicken carcass for subsequent handling, which overcomes some common problems in image segmentation, feature identification, and pose estimation for classifying, identifying, and locating poultry meat products. First, artificial color contrast and principal component analysis are optimized to enhance the contrast between similar colors to extract targeted regions more effectively. Second, the boundaries of poultry products are mostly smooth curves that are proposed as features for object recognition. Third, image feature points are found using template matching, from which the object (size, location, and pose) can be accurately determined. The algorithm simplifies the complex edges of a poultry product into a finite number of arc-centers, which greatly improves the efficiency/accuracy of the matching for part-presentation. Evaluated with 100 randomly posed samples, the algorithm has a success rate of 93%. The 7% failures were primarily due to missing points, which can be eliminated by incorporating more template points.

Fast prediction in marmoset reach-to-grasp movements for dynamic prey

Primates have evolved sophisticated, visually guided reaching behaviors for interacting with dynamic objects, such as insects, during foraging.1,2,3,4,5 Reaching control in dynamic natural conditions requires active prediction of the target's future position to compensate for visuo-motor processing delays and to enhance online movement adjustments.6,7,8,9,10,11,12 Past reaching research in non-human primates mainly focused on seated subjects engaged in repeated ballistic arm movements to either stationary targets or targets that instantaneously change position during the movement.13,14,15,16,17 However, those approaches impose task constraints that limit the natural dynamics of reaching. A recent field study in marmoset monkeys highlights predictive aspects of visually guided reaching during insect prey capture among wild marmoset monkeys.5 To examine the complementary dynamics of similar natural behavior within a laboratory context, we developed an ecologically motivated, unrestrained reach-to-grasp task involving live crickets. We used multiple high-speed video cameras to capture the movements of common marmosets (Callithrix jacchus) and crickets stereoscopically and applied machine vision algorithms for marker-free object and hand tracking. Contrary to estimates under traditional constrained reaching paradigms, we find that reaching for dynamic targets can operate at incredibly short visuo-motor delays around 80ms, rivaling the speeds that are typical of the oculomotor systems during closed-loop visual pursuit.18 Multivariate linear regression modeling of the kinematic relationships betweenthe hand and cricket velocity revealed that predictions of the expected future location can compensate for visuo-motor delays during fast reaching. These results suggest a critical role of visual prediction facilitatingonline movement adjustments for dynamic prey.

Discovery and Classification of Defects on Facing Brick Specimens Using a Convolutional Neural Network

In recent years, visual automatic non-destructive testing using machine vision algorithms has been widely used in industry. This approach for detecting, classifying, and segmenting defects in building materials and structures can be effectively implemented using convolutional neural networks. Using intelligent systems in the initial stages of manufacturing can eliminate defective building materials, prevent the spread of defective products, and detect the cause of specific damage. In this article, the solution to the problem of building elements flaw detection using the computer vision method was considered. Using the YOLOv5s convolutional neural network for the detection and classification of various defects of the structure, the appearance of finished products of facing bricks that take place at the production stage is shown during technological processing, packaging, transportation, or storage. The algorithm allows for the detection of foreign inclusions, broken corners, cracks, and color unevenness, including the presence of rust spots. To train the detector, our own empirical database of images of facing brick samples was obtained. The set of training data for the neural network algorithm for discovering defects and classifying images was expanded by using our own augmentation algorithm. The results show that the developed YOLOv5s model has a high accuracy in solving the problems of defect detection: mAP0.50 = 87% and mAP0.50:0.95 = 72%. It should be noted that the use of synthetic data obtained by augmentation makes it possible to achieve a good generalizing ability from the algorithm, it has the potential to expand visual variability and practical applicability in various shooting conditions.

A Fast and Low-Power Detection System for the Missing Pin Chip Based on YOLOv4-Tiny Algorithm.

In the current chip quality detection industry, detecting missing pins in chips is a critical task, but current methods often rely on inefficient manual screening or machine vision algorithms deployed in power-hungry computers that can only identify one chip at a time. To address this issue, we propose a fast and low-power multi-object detection system based on the YOLOv4-tiny algorithm and a small-size AXU2CGB platform that utilizes a low-power FPGA for hardware acceleration. By adopting loop tiling to cache feature map blocks, designing an FPGA accelerator structure with two-layer ping-pong optimization as well as multiplex parallel convolution kernels, enhancing the dataset, and optimizing network parameters, we achieve a 0.468 s per-image detection speed, 3.52 W power consumption, 89.33% mean average precision (mAP), and 100% missing pin recognition rate regardless of the number of missing pins. Our system reduces detection time by 73.27% and power consumption by 23.08% compared to a CPU, while delivering a more balanced boost in performance compared to other solutions.

Experimental assessment and semi empirical estimation of frost accretion—A case study on a spine-finned inverted-V tube array evaporator

Present work focuses on frost accretion in a spine-finned inverted-V tube array evaporator. An experimental evaluation was performed using a standard issue, vertical top-mount, 18 cubic feet, 0.5 m3, refrigerator. Evaporator temperature distribution, inner airflow velocity, and relative humidity were measured to account for convective phenomena influencing frost distribution. Frost formation and accretion on the surface of the evaporator were visualized using thermal and microscopic imagery. The images were processed using a machine vision algorithm to measure frost thickness. Complementarily, frost density and vapor mass transfer were computed using available correlations. An estimation function was derived from the compiled data using a semi empirical approach, i.e., direct measurements and thermophysical substance properties. The resulting mathematical expression estimated the frost accretion rate within an error expectancy, RMSE, of 0.1479 and displayed a goodness-of-fit, R-Squared, of 0.9029. Based on these results, semi empirical estimation, is proposed as a viable approach to construct adequate limits for new predictions, vis-à-vis evaporator performance, ultimately reducing appliance energy consumption via implementing more effective control strategies regarding internal defrosting.

A CMOS Image Readout Circuit with On-Chip Defective Pixel Detection and Correction.

Images produced by CMOS sensors may contain defective pixels due to noise, manufacturing errors, or device malfunction, which must be detected and corrected at early processing stages in order to produce images that are useful to human users and image-processing or machine-vision algorithms. This paper proposes a defective pixel detection and correction algorithm and its implementation using CMOS analog circuits, which are integrated with the image sensor at the pixel and column levels. During photocurrent integration, the circuit detects defective values in parallel at each pixel using simple arithmetic operations within a neighborhood. At the image-column level, the circuit replaces the defective pixels with the median value of their neighborhood. To validate our approach, we designed a 128×128-pixel imager in a 0.35μm CMOS process, which integrates our defective-pixel detection/correction circuits and processes images at 694 frames per second, according to post-layout simulations. Operating at that frame rate, our proposed algorithm and its CMOS implementation produce better results than current state-of-the-art algorithms: it achieves a Peak Signal to Noise Ratio (PSNR) and Image Enhancement Factor (IEF) of 45 dB and 198.4, respectively, in images with 0.5% random defective pixels, and a PSNR of 44.4 dB and IEF of 194.2, respectively, in images with 1.0% random defective pixels.

Managing the cyber-physical security for unmanned aerial vehicles used in perimeter surveillance

This paper discusses the vulnerabilities of perimeter surveillance unmanned aerial vehicles to cyber-physical security threats and discusses some approaches to manage them, as most cyber threats to the UAVs coming through their onboard wireless transceiver, we are suggesting an Antennas propagation type that limit the vector of the threat, also the importance of vulnerabilities scanner to evaluate the system risks. And addressing the limited energy and computation power resources onboard, a computation efficient onboard encryption method is proposed, and a sign cutting machine vision algorithm to provide warning of suspicious activity detected on interrupted surveillance imagery. The focus of this analysis is to manage the vulnerabilities of the system during both its operation time and its standby time, this is done by performing a checklist of computerized tests periodically and comparing two or more results to discover any unexpected changes. We showed that the onboard resources can be utilized more efficiently securing itself and the system from possible intrusion, however, some heavy calculation tasks still need to be performed on the ground control station which is causing some latency problem to such time sensitive operation. We have demonstrated that image registration techniques have produced useful results when applied to analyze the differences between two scene images.

Approaches to Creating a Driver Decision Support System for Digital Analysis of Railway Infrastructure Based on Machine Learning and Machine Vision Algorithms

The paper considers the issues of creating a driver decision support system for digital analysis of the railway infrastructure based on machine learning and machine vision algorithms, which will take into account and analyse the given traffic schedule, infrastructure capabilities, dispatch centre teams, statuses of the nearest traffic participants for unmanned safe control of electric rolling stock. A detailed review of existing control systems in railway transport is made, which are based on technical vision.

Multiparameter mobile blood analysis for complete blood count using contrast-enhanced defocusing imaging and machine vision.

Blood analysis through complete blood count is the most basic medical test for disease diagnosis. Conventional blood analysis requires bulky and expensive laboratory facilities and skilled technicians, limiting the universal medical use of blood analysis outside well-equipped laboratory environments. Here, we propose a multiparameter mobile blood analyzer combined with label-free contrast-enhanced defocusing imaging (CEDI) and machine vision for instant and on-site diagnostic applications. We designed a low-cost and high-resolution miniature microscope (size: 105 mm × 77 mm × 64 mm, weight: 314 g) that comprises a pair of miniature aspheric lenses and a 415 nm LED for blood image acquisition. The analyzer, adopting CEDI, can obtain both the refractive index distributions of the white blood cell (WBC) and hemoglobin spectrophotometric information, enabling the analyzer to supply rich blood parameters, including the five-part WBC differential count, red blood cell (RBC) count, and mean corpuscular hemoglobin (MCH) quantification with machine vision algorithms and the Lambert-Beer law. We have shown that our assay can analyze a blood sample within 10 minutes without complex staining, and measurements (30 samples) from the analyzer have a strong linear correlation with clinical reference values (significance level of 0.0001). This study provides a miniature, light weight, low-cost, and easy-to-use blood analysis technique that overcomes the challenge of simultaneously realizing FWD count, RBC count, and MCH analysis using a mobile device and has great potential for integrated surveillance of various epidemic diseases, including coronavirus infection, invermination, and anemia, especially in low- and middle-income countries.