Machine Vision Intelligence Research Articles

The Exploration and Research of AI+ Intelligent Machine Vision in Practice Teaching

The rapid development and gradual perfection of artificial intelligence (AI) and intelligent machine vision technology have promoted the expansion of their application fields. And at this stage, AI+ intelligent machine vision has been widely used in practical teaching. Moreover, because artificial intelligence (AI) and intelligent machine vision technology involve many knowledge subjects, it is necessary to explore and study them effectively in order to give full play to their effectiveness in practical teaching.

Integration and performance analysis of artificial intelligence and computer vision based on deep learning algorithms

This paper focuses on the analysis of the application effectiveness of the integration of deep learning and computer vision technologies. Deep learning achieves a historic breakthrough by constructing hierarchical neural networks, enabling end-to-end feature learning and semantic understanding of images. The successful experiences in the field of computer vision provide strong support for training deep learning algorithms. The tight integration of these two fields has given rise to a new generation of advanced computer vision systems, significantly surpassing traditional methods in tasks such as machine vision image classification and object detection. In this paper, typical image classification cases are combined to analyze the superior performance of deep neural network models while also pointing out their limitations in generalization and interpretability, proposing directions for future improvements. Overall, the efficient integration and development trend of deep learning with massive visual data will continue to drive technological breakthroughs and application expansion in the field of computer vision, making it possible to build truly intelligent machine vision systems. This deepening fusion paradigm will powerfully promote unprecedented tasks and functions in computer vision, providing stronger development momentum for related disciplines and industries.

Neuromorphic Vision Sensor driven by Ferroelectric HfAlO

The rapid progress in bioinspired machine vision supported by intelligent computing has attracted attention of the modern electronics due to their wide applications in image recognition, artificial intelligence, and medical applications. The ferroelectric materials are considered as most reliable for the fabrication of these devices due to their nonvolatile and dense memory capacity as well as stability. Here, we report the neuromorphic vision sensor (NVS) based on a self-grown ferroelectric interfacial HfAlO layer. The unique properties of graphene (Gr) and silicon are integrated through the ferroelectric layer to fabricate the NVS, which can mimic the human eye system. The sensor efficiently demonstrates the basic neural functions including pair-pulsed facilitation, depression, inhibition, and excitation under electric as well as optical stimuli. The optical spike rate (2 Hz–10 Hz) as well as the amplitude (0.5 μW/cm2-1.5 μW/cm2) dependent plasticity, exhibited excellent facilitation and depression index of 1.62% and −1.86% with the minimum energy consumption of only 3.5 fJ per spike. The high endurance of 1 × 107 cycles confirm its reliability and data security. Our device with excellent features of optical signal detection, information processing, and data storage with its simple structure can be efficiently utilized as an image sensor for robotics and intelligent machine vision systems.

Research on Concrete Compressive Strength Detection Technology Based on Intelligent Machine Vision

Abstract Concrete is the most common and important building material nowadays. Its compressive strength plays a crucial role in the result and safety of the building. To improve the efficiency of concrete compressive strength detection, this study combines intelligent machine vision technology to design a concrete compressive strength detection system. The features of concrete are extracted using the edge detection method. Then the extracted features are classified using the random forest method to complete the identification and localization of concrete. Based on this basis, the compressive strength of concrete is calculated and detected based on the conversion relationship between uniaxial compressive strength and point load strength. Finally, after testing the performance of the system, the practical effects of the system are examined. According to the results, the system’s detection rate is between 0.058 and 0.072 seconds, and the recognition accuracy and classification accuracy of the four different types of concrete detection exceed 80%. The relative error values for the detected compressive strength were 5.87% and 3.52%, respectively, and they passed the compressive strength detection of retardation diagrams in complex situations. The excellent performance of this study in real concrete detection meets the demand for concrete compressive detection in reality.

Integrating intelligent machine vision techniques to advance precision manufacturing: a comprehensive survey in the context of mechatronics and beyond

Integrating intelligent machine vision techniques to advance precision manufacturing: a comprehensive survey in the context of mechatronics and beyond

Parallel perception of visual motion using light-tunable memory matrix.

Parallel perception of visual motion is of crucial significance to the development of an intelligent machine vision system. However, implementing in-sensor parallel visual motion perception using conventional complementary metal-oxide semiconductor technology is challenging, because the temporal and spatial information embedded in motion cannot be simultaneously encoded and perceived at the sensory level. Here, we demonstrate the parallel perception of diverse motion modes at the sensor level by exploiting light-tunable memory matrix in a van der Waals (vdW) heterostructure array. The optoelectronic characteristics of gate-tunable photoconductivity and light-tunable memory matrix enable devices in the array to realize simultaneous encoding and processing of incoming spatiotemporal light pattern. Furthermore, we implement a visual motion perceptron with the array capable of deciphering multiple motion parameters in parallel, including direction, velocity, acceleration, and angular velocity. Our work opens up a promising venue for the realization of an intelligent machine vision system based on in-sensor motion perception.

Intellectualized Visualization of Single-Particle Raman Spectra for Sensitive Detection and Simultaneous Multianalysis of Heavy Metal Ions.

Easy-to-use, reliable, and real-time methods for detecting heavy metal ion contamination are urgently required, which is a primary concern for water pollution control and human health. However, present methods for this aim are still unable to achieve simultaneous multianalysis for complex real sample detection. Herein, an intellectualized vision-based single-nanoparticle Raman imaging strategy combined with ion-responsive functional nucleic acids (FNAs) was proposed to address these issues. We reported a correspondence between the concentration of the analytes and the density of particles (DOP) of specifically captured nanoparticles to achieve sensitive detection and simultaneous multianalysis of heavy metal ions. The specific detection of Pb2+ (Hg2+) was obtained with a detection linear range from 100 pM to 100 nM (from 500 fM to 100 nM) and limit of detections low to 1 pM (100 fM), with the advantages of good specificity, excellent homogeneity, and reproducibility. Furthermore, the differentiation of different heavy metal ions (Pb2+/Hg2+) was achieved, i.e., the simultaneous multianalysis, based on Raman imaging of the single particle and intelligent machine vision method. Finally, the Raman imaging assay was utilized for real sample analysis, and it provided a powerful and reliable tool for detecting trace Pb2+/Hg2+ in real water samples and facilitated the portable on-site monitoring of heavy metal ions.

Multi-wavelength diffractive neural network with the weighting method.

Recently, the diffractive deep neural network (D2NN) has demonstrated the advantages to achieve large-scale computational tasks in terms of high speed, low power consumption, parallelism, and scalability. A typical D2NN with cascaded diffractive elements is designed for monochromatic illumination. Here, we propose a framework to achieve the multi-wavelength D2NN (MW-D2NN) based on the method of weight coefficients. In training, each wavelength is assigned a specific weighting and their output planes construct the wavelength weighting loss function. The trained MW-D2NN can implement the classification of images of handwritten digits at multi-wavelength incident beams. The designed 3-layers MW-D2NN achieves a simulation classification accuracy of 83.3%. We designed a 1-layer MW-D2NN. The simulation and experiment classification accuracy are 71.4% and 67.5% at RGB wavelengths. Furthermore, the proposed MW-D2NN can be extended to intelligent machine vision systems for multi-wavelength and incoherent illumination.

Development of a Finite Element Model to Analyze the Lateral Drift of Autonomous Vehicles and Its Impacts on Rutting Depth on Flexible Pavements Under Mixed Traffic Scenarios

The continuous development of technologies, including intelligent networking, machine vision, and new energy vehicles, has made autonomous driving vehicles a research focus. Autonomous vehicles significantly affect road performance under mixed traffic flow conditions with their excellent lateral control ability and driving stability. A finite element model is established to analyze the impact of these modes on pavement deformation under maximum air temperature in Nanjing, China. Based on the simulation results, it has been shown that lateral control of autonomous vehicles significantly reduces pavement deformation. The proportion of autonomous vehicles affects the difference in distribution modes, and the uniform distribution mode is the most favorable for horizontal distribution. Additionally, the edge lane experiences less pavement deformation than the center lane and varies more when the distribution pattern changes. By comparing the deformation of the center and edge lanes, the edge lane experiences a significant reduction in deformation under the uniform distribution mode, with a maximum reduction of 73.7%. Comparing the pavement deformation of center and edge lanes under normal distribution mode and uniform distribution mode shows that safe solutions for the lateral distribution of autonomous vehicles depend on the proportion of autonomous vehicles. The optimal driving speed for autonomous vehicles is 80 km/h under the uniform distribution model, while under the normal distribution model it is 90 km/h or 110 km/h. These findings provide a theoretical basis for the lateral control of autonomous vehicles in the future and propose safe solutions for the lateral distribution of autonomous vehicles.

Self-powered and broadband opto-sensor with bionic visual adaptation function based on multilayer γ-InSe flakes

Visual adaptation that can autonomously adjust the response to light stimuli is a basic function of artificial visual systems for intelligent bionic robots. To improve efficiency and reduce complexity, artificial visual systems with integrated visual adaptation functions based on a single device should be developed to replace traditional approaches that require complex circuitry and algorithms. Here, we have developed a single two-terminal opto-sensor based on multilayer γ-InSe flakes, which successfully emulated the visual adaptation behaviors with a new working mechanism combining the photo-pyroelectric and photo-thermoelectric effect. The device can operate in self-powered mode and exhibit good human-eye-like adaptation behaviors, which include broadband light-sensing image adaptation (from ultraviolet to near-infrared), near-complete photosensitivity recovery (99.6%), and synergetic visual adaptation, encouraging the advancement of intelligent opto-sensors and machine vision systems.

Massively parallel universal linear transformations using a wavelength-multiplexed diffractive optical network

We report deep learning-based design of a massively parallel broadband diffractive neural network for all-optically performing a large group of arbitrarily-selected, complex-valued linear transformations between an input and output field-of-view, each with N_i and N_o pixels, respectively. This broadband diffractive processor is composed of N_w wavelength channels, each of which is uniquely assigned to a distinct target transformation. A large set of arbitrarily-selected linear transformations can be individually performed through the same diffractive network at different illumination wavelengths, either simultaneously or sequentially (wavelength scanning). We demonstrate that such a broadband diffractive network, regardless of its material dispersion, can successfully approximate N_w unique complex-valued linear transforms with a negligible error when the number of diffractive neurons (N) in its design matches or exceeds 2 x N_w x N_i x N_o. We further report that the spectral multiplexing capability (N_w) can be increased by increasing N; our numerical analyses confirm these conclusions for N_w > 180, which can be further increased to e.g., ~2000 depending on the upper bound of the approximation error. Massively parallel, wavelength-multiplexed diffractive networks will be useful for designing high-throughput intelligent machine vision systems and hyperspectral processors that can perform statistical inference and analyze objects/scenes with unique spectral properties.

Automatic fabric defect detection employing deep learning

A major issue for fabric quality inspection is in the detection of defaults, it has become an extremely challenging goal for the textile industry to minimize costs in both production and quality inspection. The quality inspection is currently done manually by professionals; hence the need for the implementation of a fast, powerful, robust, and intelligent machine vision system in order to achieve high global quality, uniformity, and consistency of fabrics and to increase productivity. Consequently, the automatic inspection control process can improve productivity and enhance product quality. This article describes the approach used in developing a convolutional neural network for identifying fabric defects from input images of fabric surfaces. The proposed neural network is a pre-trained convolutional model ‘DetectNet’, it was adapted to be more efficient to the fabric image feature extraction. The developed model is capable of successfully distinguishing between defective fabric and non-defective with 93% accuracy for the first model and 96% for the second model.

Diagnosis of grape leaf diseases using automatic K-means clustering and machine learning

Plant diseases often reduce crop yield and product quality; therefore, plant disease diagnosis plays a vital role in farmers’ management decisions. Visual crop inspections by humans are time-consuming and challenging tasks and, practically, can only be performed in small areas at a given time, especially since many diseases have similar symptoms. An intelligent machine vision monitoring system for automatic inspection can be a great help for farmers in this regard. Although many algorithms have been introduced for plant disease diagnosis in recent years, a simple method relying on minimal information from the images is of interest for field conditions. In this study, a novel image processing algorithm and multi-class support vector machine (SVM) were used to diagnose and classify grape leaf diseases, i.e., black measles, black rot, and leaf blight. The area of disease symptoms was separated from the healthy parts of the leaf utilizing K-means clustering automatically, and then the features were extracted in three color models, namely RGB, HSV, and l*a*b. As an efficient classification method, SVM was used in this study, where principal component analysis (PCA) was performed for feature dimension reduction. Finally, the most important features were selected by the relief feature selection. Gray-level co-occurrence matrix (GLCM) features resulted in an accuracy of 98.71%, while feature dimension reduction using PCA resulted in an accuracy of 98.97%. The proposed method was compared with two deep learning methods, i.e., CNN and GoogleNet, which achieved classification accuracies of 86.82% and 94.05%, respectively, while the processing time for the proposed method was significantly shorter than those of these models.

UAV Intelligent Control Based on Machine Vision and Multiagent Decision-Making

In order to improve the effect of UAV intelligent control, this paper will improve machine vision technology. Moreover, this paper adds scale information on the basis of the LSD algorithm, uses the multiline segment standard to merge these candidate line segments for intelligent recognition, and uses the LSD detection algorithm to improve the operating efficiency of the UAV control system and reduce the computational complexity. In addition, this paper combines machine vision technology and multiagent decision-making technology for UAV intelligent control and builds an intelligent control system, which uses intelligent machine vision technology for recognition and multiagent decision-making technology for motion control. The research results show that the UAV intelligent control system based on machine vision and multiagent decision-making proposed in this paper can achieve reliable control of UAVs and improve the work efficiency of UAVs.

Design of robot automatic navigation under computer intelligent algorithm and machine vision

This work aims to explore the robot automatic navigation model under computer intelligent algorithms and machine vision, so that mobile robots can better serve all walks of life. In view of the current situation of high cost and poor work flexibility of intelligent robots, this work innovatively researches and improves the image processing algorithm and control algorithm. In the navigation line edge detection stage, aiming at the low efficiency of the traditional ant colony algorithm, the Canny algorithm is combined to improve it, and a Canny-based ant colony algorithm is proposed to detect the trajectory edge. In addition, the Single Shot MultiBox Detector (SSD) algorithm is adopted to detect obstacles in the navigation trajectory of the robot. The performance is analyzed through simulation. The results show that the navigation accuracy of the Canny-based ant colony algorithm proposed in this work is basically stable at 89.62%, and its running time is the shortest. Further analysis of the proposed SSD neural network through comparison with other neural networks suggests that its feature recognition accuracy reaches 92.90%. The accuracy is at least 3.74% higher versus other neural network algorithms, the running time is stable at about 37.99 s, and the packet loss rate is close to 0. Therefore, the constructed mobile robot automatic navigation model can achieve high recognition accuracy under the premise of ensuring error. Moreover, the data transmission effect is ideal. It can provide experimental basis for the later promotion and adoption of mobile robots in various fields.

A video processing and machine vision-based automatic analyzer to determine sequentially total suspended and settleable solids in wastewater

The monitoring of total suspended (TSS) and settleable (SetS) solids in wastewater is essential to maintain the quality parameters for aquatic biota because they can transport pollutants and block light penetration. Determining them by their respective reference methods, however, is laborious, expensive, and time consuming. To overcome this, we developed a new analytical instrument called Solids in Wastewater's Machine Vision-based Automatic Analyzer (SWAMVA), which is equiped with an automatic sampler and a software for real-time digital movie capture to quantify sequentially the TSS and SetS contents in wastewater samples. The machine vision algorithm (MVA) coupled with the Red color plane (derived from color histograms in the Red-Green-Blue (RGB) system) showed the best prediction results with R2 of 0.988 and 0.964, and relative error of prediction (REP) of 6.133 and 9.115% for TSS and SetS, respectively. The constructed models were validated by Analysis of Variance (ANOVA), and the accuracy and precision of the predictions by the t- and F-tests, respectively, at a 0.05 significance level. The elliptical joint confidence region (EJCR) test confirmed the accuracy, while the coefficient of variation (CV) of 6.529 and 10.908% confirmed the good precisions, respectively. Compared with the reference method (Standard Methods For the Examination of Water and Wastewater), the proposed method reduced the analysis volume from 1.5 L to just 15 mL and the analysis time from 12 h to 24 s per sample. Therefore, SWAMVA can be considered an important alternative to the determination of TSS and SetS in wastewater as an automatic, fast, and low-cost analytical tool, following the principles of Green Chemistry and exploiting Industry 4.0 features such as intelligent processing, miniaturization, and machine vision.

Intelligent machine vision model for building architectural style classification based on deep learning

Intelligent machine vision model for building architectural style classification based on deep learning

Intelligent machine vision model for building architectural style classification based on deep learning

This paper presents an intelligent model for building architectural style classification. Image classification of architectural style is challenging to traditional machine vision methods. The main challenge in these systems is the feature extraction phase as there are many visual features in these styles that need to be extracted, refined and optimised. All these operations are done at the researcher discretion in traditional Machine Learning (ML) models. The advancements of ML to Deep Learning (DL) made automation of all the challenging operations possible. We constructed a machine vision model based on DL to investigate the effectiveness of DL in the classification problem at hand. A publicly available annotated data set was utilised to train and validate the proposed model. The data set consists of more than 5000 images of eight different architectural styles. The experimental results showed that the proposed model is reliable as it produced a classification accuracy of 95.44%.

Face recognition robot system based on intelligent machine vision image recognition

Face recognition robot system based on intelligent machine vision image recognition

Intelligent Machine Vision Model for Defective Product Inspection Based on Machine Learning

Quality control is one of the industrial tasks most susceptible to be improved by implementing technological innovations. As an innovative technology, machine vision enables reliable and fast 24/7 inspections and helps producers to improve the efficiency of manufacturing operations. The accessible data by vision equipment will be used to identify and report defective products, understand the causes of deficiencies and allow rapid and efficient intervention in smart factories. From this perspective, the proposed machine vision model in this paper combines the identification of defective products and the continuous improvement of manufacturing processes by predicting the most suitable parameters of production processes to obtain a defect-free item. The suggested model exploits all generated data by various integrated technologies in the manufacturing chain, thus meeting the requirements of quality management in the context of Industry 4.0, based on predictive analysis to identify patterns in data and suggest corrective actions to ensure product quality. In addition, a comparative study between several machine learning algorithms, both for product classification and process improvement models, is performed in order to evaluate the designed system. The results of this study show that the proposed model largely meets the requirements for the proper implementation of these techniques.