Image Extraction Method Research Articles

A method for extracting corneal reflection images from multiple eye images

The incident light reflected from the cornea is rich in information about the human surroundings, and these reflected rays are imaged by the camera, which can be used for research on human consciousness and gaze analysis, and produce certain help in the fields of psychology, human computer interaction and disease diagnosis. However, limited by the low corneal reflection ability, when a high-definition camera captures corneal reflecting rays, a large amount of color and texture interference from the iris can seriously contaminate the corneal reflection images, resulting in low usability and ubiquity of corneal reflection images. In this paper, we propose a corneal reflection image extraction method with multiple eye images as input. We align the iris regions of multiple eye images with the help of iris localization method, and by comparing multiple iris regions, we obtain the complementary iris regions, so that the iris interference in the corneal reflection region can be stripped completely. A large number of experiments have demonstrated that our work can effectively mitigate iris interference and effectively improve the quality of corneal reflection images.

Recommender-based bone tumour classification with radiographs—a link to the past

ObjectivesTo develop an algorithm to link undiagnosed patients to previous patient histories based on radiographs, and simultaneous classification of multiple bone tumours to enable early and specific diagnosis.Materials and methodsFor this retrospective study, data from 2000 to 2021 were curated from our database by two orthopaedic surgeons, a radiologist and a data scientist. Patients with complete clinical and pre-therapy radiographic data were eligible. To ensure feasibility, the ten most frequent primary tumour entities, confirmed histologically or by tumour board decision, were included. We implemented a ResNet and transformer model to establish baseline results. Our method extracts image features using deep learning and then clusters the k most similar images to the target image using a hash-based nearest-neighbour recommender approach that performs simultaneous classification by majority voting. The results were evaluated with precision-at-k, accuracy, precision and recall. Discrete parameters were described by incidence and percentage ratios. For continuous parameters, based on a normality test, respective statistical measures were calculated.ResultsIncluded were data from 809 patients (1792 radiographs; mean age 33.73 ± 18.65, range 3–89 years; 443 men), with Osteochondroma (28.31%) and Ewing sarcoma (1.11%) as the most and least common entities, respectively. The dataset was split into training (80%) and test subsets (20%). For k = 3, our model achieved the highest mean accuracy, precision and recall (92.86%, 92.86% and 34.08%), significantly outperforming state-of-the-art models (54.10%, 55.57%, 19.85% and 62.80%, 61.33%, 23.05%).ConclusionOur novel approach surpasses current models in tumour classification and links to past patient data, leveraging expert insights.Clinical relevance statementThe proposed algorithm could serve as a vital support tool for clinicians and general practitioners with limited experience in bone tumour classification by identifying similar cases and classifying bone tumour entities.Key Points• Addressed accurate bone tumour classification using radiographic features.• Model achieved 92.86%, 92.86% and 34.08% mean accuracy, precision and recall, respectively, significantly surpassing state-of-the-art models.• Enhanced diagnosis by integrating prior expert patient assessments.Graphical abstract

Exploration of the Integration and Innovation of Digital Art Elements and Traditional Cultural and Creative Designs

Abstract Combined with the modern development form of traditional cultural and creative design, this paper proposes to use digital art elements to assist traditional cultural and creative design. The symbols of cultural and creative design are extracted using digital-assisted design technology, and the grayscale image extraction method is chosen to process the images of cultural and creative design. Then, combined with the YOLO v3 algorithm, CMT algorithm, and spatial mapping to realize three-dimensional tracking of real objects and real-time tracking interactions in mixed reality, adding digitally assisted experience of cultural symbols. With that, the theoretical model of digital art elements and traditional cultural and creative design is designed, and the procedural method of the entire cultural and creative design is analyzed. Online shopping consumer evaluations are selected, high-frequency words are classified, and consumers’ emotional tendencies are analyzed as well. The comprehensive scores of digital art and cultural and creative products are calculated using sense-quality evaluations. Consumers’ positive and negative emotions accounted for 75.72% and 17.49%, respectively, with the general negative emotion field accounting for the highest proportion of 10.54% and the medium and high negative emotion fields accounting for a lower proportion. It can be seen that the response of digital art elements to the cultural and creative design market is better, and its innovative design has a certain consumer market.

Assessment of satellite-derived shorelines automatically extracted from Sentinel-2 imagery using SAET

The definition of the shoreline position from satellite imagery is of great interest among coastal monitoring techniques. Understanding the reality mapped by the resulting shorelines and defining their accuracy is of paramount importance. The assessment described in this paper constitutes a validation of the shorelines obtained by using the novel tool SAET (Shoreline Analysis and Extraction Tool) for automatic shoreline extraction. The resulting shorelines applying the different parameters available in SAET are assessed in 9 test sites with diverse morphology and oceanographic conditions along the Atlantic European and Western Mediterranean coasts. The reference data is obtained along large coastal segments (covering up to about 240 km) from nearly coincident very high-resolution satellite images.Different image processing levels and extraction methods have been tested, showing their key role in the accuracy of shoreline position. When defining the approximate shoreline position the Automated Water Extraction Index for images without shadows (AWEInsh) with a 0 threshold generally constitutes the best segmentation method. In turn, the employment of the mathematical morphological operations of dilation or erosion considerably improves the results in certain coastal typologies. On the contrary, the employment of atmospherically-corrected images has a smaller influence on the accuracy of the SDSs.Results support the idea that the magnitude of the errors is strongly related to the specific coastal conditions- In general, the lowest errors appear in low-energetic microtidal sites, contrary to the energetic and mesotidal coasts with gentle slopes.The shoreline errors range between 3.7 m and 13.5 m RMSE (root-mean-square error) among the different coastal types when selecting the most appropriate extraction parameters. The shoreline position identified with SAET shows a similar or better accuracy to that obtained by other tools.

An AR Map Virtual–Real Fusion Method Based on Element Recognition

The application of AR to explore augmented map representation has become a research hotspot due to the growing application of AR in maps and geographic information in addition to the rising demand for automated map interpretation. Taking the AR map as the research object, this paper focuses on AR map tracking and registration and the virtual–real fusion method based on element recognition. It strives to establish a new geographic information visualization interface and application model. AR technology is applied to the augmented representation of 2D planar maps. A step-by-step identification and extraction method of unmarked map elements are designed and proposed based on the analysis of the characteristics of planar map elements. This method combines the spatial and attribute characteristics of point-like elements and line-like elements, extracts the color, geometric features, and spatial distribution of map elements through computer vision methods, and completes the identification and automatic extraction of map elements. The multi-target image recognition and extraction method based on template and contour matching, and the line element recognition and extraction method based on color space and area growth are introduced in detail. Then, 3D tracking and registration is used to realize the unmarked tracking and registration of planar map element images, and the AR map virtual–real fusion algorithm is proposed. The experimental results and results of an analysis of stepwise identification and extraction of unmarked map elements and map virtual–real fusion reveal that the stepwise identification of unmarked map elements and map model virtual–real fusion studied in this paper is effective. Through the analysis of map element step-by-step recognition efficiency and recognition rate, it is proved that the element step-by-step method in this paper is fast, its recognition efficiency meets the AR real-time requirements, and its recognition accuracy is high.

Transfer and Unsupervised Learning: An Integrated Approach to Concrete Crack Image Analysis

The detection of cracks in concrete structures is crucial for the assessment of their structural integrity and safety. To this end, detection with deep neural convolutional networks has been extensively researched in recent years. Despite their success, these methods are limited in classifying concrete as cracked or non-cracked and disregard other characteristics, such as the severity of the cracks. Furthermore, the classification process can be affected by various sources of interference and noise in the images. In this paper, an integrated methodology for analysing concrete crack images is proposed using transfer and unsupervised learning. The method extracts image features using pre-trained networks and groups them based on similarity using hierarchical clustering. Three pre-trained networks are used for this purpose, with Inception v3 performing the best. The clustering results show the ability to divide images into different clusters based on image characteristics. In this way, various clusters are identified, such as clusters containing images of obstruction, background debris, edges, surface roughness, as well as cracked and uncracked concrete. In addition, dimensionality reduction is used to further separate and visualise the data, making it easier to analyse clustering results and identify misclassified images. This revealed several mislabelled images in the dataset used in this study. Additionally, a correlation was found between the principal components and the severity of cracks and surface imperfections. The results of this study demonstrate the potential of unsupervised learning for analysing concrete crack image data to distinguish between noisy images and the severity of cracks, which can provide valuable information for building more accurate predictive models.

High-Accuracy Three-Dimensional Deformation Measurement System Based on Fringe Projection and Speckle Correlation.

Fringe projection profilometry (FPP) and digital image correlation (DIC) are widely applied in three-dimensional (3D) measurements. The combination of DIC and FPP can effectively overcome their respective shortcomings. However, the speckle on the surface of an object seriously affects the quality and modulation of fringe images captured by cameras, which will lead to non-negligible errors in the measurement results. In this paper, we propose a fringe image extraction method based on deep learning technology, which transforms speckle-embedded fringe images into speckle-free fringe images. The principle of the proposed method, 3D coordinate calculation, and deformation measurements are introduced. Compared with the traditional 3D-DIC method, the experimental results show that this method is effective and precise.

Facial emotions recognition using local monotonic pattern and grey level co‐occurrence matrices images aided development

AbstractIn this article, local monotonic pattern (LMP) paired with grey level co‐occurrence matrix (GLCM) methods are suggested to identify facial emotions with a high identification rate even when the face pictures are rotated. The proposed method extracts image features using the properties of the LMP algorithm and features extracted from the GLCM, which are then fed into the Support Vector Machine (SVM) process that reduces the dimensionality of the features vector and classifies the output into different facial expressions or emotions. The SVM performance rate is then compared to the K‐nearest neighbour approach (KNN) to see which method produces the best facial emotion identification and categorization. The study identified facial emotions in the images using advanced algorithms of GLCM and LMP models to be compared. As a result, the accuracy of SVM and KNN was utilized to determine the method's usefulness in classification using the application of MATLAB. A result of more than 93% was achieved using the SVM method compared with 89.4% using the KNN for the recognition process. The study also demonstrated that this approach would lead to more classification outcomes if the LMP and GLCM are combined with an edge‐based technique yielding a new method that is more efficient and more effective.

Facial emotions recognition using local monotonic pattern and gray level co-occurrence matrices plant leaf images aided agriculture development

In this article, Local Monotonic Pattern (LMP) paired with Gray Level Co-occurrence Matrix (GLCM) methods are suggested in this study to identify facial emotions with a high identification rate even when the face pictures are rotated. The proposed method extracts image features using the properties of the LMP algorithm and features extracted from the GLCM, which are then fed into the Support Vector Machine (SVM) process that reduces the dimensionality of the features vector and classifies the output into different facial expressions or emotions. The SVM performance rate is then compared to the K-Nearest Neighbor approach (KNN) to see which method produces the best facial emotion identification and categorization. Consequently, the study identified facial emotions in the images using advanced algorithms of GLCM and LMP models to be compared with. As a result, the accuracy of SVM and KNN was utilized to determine the method's usefulness in classification using the application of MATLAB. A result of more than 93% has been achieved using SVM method compared with 89.4% using the KNN for the recognition process. The study also demonstrated that this approach will result in a more classification outcomes if the LMP and GLCM are combined with an edge-based technique which yield in a new method that is more efficient and more effective.

3D Face Recognition Neural Network for Digital Human Resource Management

Digital human resource management can improve the organizational and operational efficiency of enterprises.In order to improve the efficiency of enterprise digital management and solve the problems of low security level and insufficient stability of 2D face recognition, we introduce 3D face recognition into the digital human resource management system. We propose a face recognition method based on a multistream convolutional neural network and local binary pattern and build a digital face recognition management system. We first build the system computer vision scene. Then a local binary mode facial expression feature extraction scheme is designed according to the depth camera image extraction method. Considering that face 3D features are easy to be missed, we build a multistream convolutional neural network to learn facial 3D features. Finally, we validate the effectiveness of the method in selecting a public face dataset. Experiments prove that our method can reach 98% face recognition accuracy, which is significantly better than other methods.

Evaluating Image Extraction Methods Over Different Types of Images

Evaluating Image Extraction Methods Over Different Types of Images

Recognition and sorting of coal and gangue based on image process and multilayer perceptron

ABSTRACT The recognition and sorting of coal and gangue are an important part of the process of reducing costs and improving production efficiency. In this paper, under the influence of long running pollution and similar background grayscale, a new method was proposed based on image process and multilayer perceptron. The robust image extraction method was used for the reverse selection edge extraction method (RS-EEM) to eliminate background interference. Moreover, a deep learning model called multilayer perceptron (MLP) was proposed to recognize coal and gangue with simple structure, which is easy to transplant with high precision and great timeliness. In order to evaluate the correctness and efficiency of the algorithm, a sorting robot of coal and gangue was built based on MLP and image processing. The experiment results showed that the images of coal and gangue can be precisely recognized based on RS-EEM under the influence of long running pollution and similar background grayscale, and the recognition accuracy can reach to 96.15% as well as grasping accuracy to 85% at 0.4 m/s conveyor belt speed.

Research on Oil Painting Image Extraction and Identification Method based on Intelligent Vision

A high-precision and high-efficiency oil painting identification method is the auxiliary basis for authenticating works, because it can improve the efficiency and credibility of oil painting identification. Therefore, an oil painting image extraction method based on intelligent vision was proposed.An intelligent visual detection model was constructed to obtain the characteristics of oil painting images.The oil painting feature fusion method based on intelligent vision was adopted to integrate the color and shape features of oil painting features, calculate oil painting feature difference coefficient and difference feature threshold, and realize oil painting image extraction by oil painting image extraction rules.The research results verified that the proposed method could effectively identify the authenticity of oil paintings. Compared with the expert identification method and the identification method based on deep learning, it can be seen that the method had the highest identification accuracy, the shortest identification time, the best anti-interference, and the remarkable identification performance, so it had a high application value.

Deep learning based algorithm (ConvLSTM) for Copy Move Forgery Detection

Protecting information from manipulation is important challenge in current days. Digital images are one of the most popular information representation. Images could be used in several fields such as military, social media, security purposes, intelligence fields, evidences in courts, and newspapers. Digital image forgeries mean adding unusual patterns to the original images that cause a heterogeneity manner in form of image properties. Copy move forgery is one of the hardest types of image forgeries to be detected. It is happened by duplicating part or section of the image then adding again in the image itself but in another location. Forgery detection algorithms are used in image security when the original content is not available. This paper illustrates a new approach for Copy Move Forgery Detection (CMFD) built basically on deep learning. The proposed model is depending on applying (Convolution Neural Network) CNN in addition to Convolutional Long Short-Term Memory (CovLSTM) networks. This method extracts image features by a sequence number of Convolutions (CNVs) layers, ConvLSTM layers, and pooling layers then matching features and detecting copy move forgery. This model had been applied to four aboveboard available databases: MICC-F220, MICC-F2000, MICC-F600, and SATs-130. Moreover, datasets have been combined to build new datasets for all purposes of generalization testing and coping with an over-fitting problem. In addition, the results of applying ConvLSTM model only have been added to show the differences in performance between using hybrid ConvLSTM and CNN compared with using CNN only. The proposed algorithm, when using number of epoch’s equal 100, gives high accuracy reached to 100% for some datasets with lowest Testing Time (TT) time nearly 1 second for some datasets when compared with the different previous algorithms.

A coarse-to-fine registration method for three-dimensional MR images.

Three-dimensional (3D) multimodal magnetic resonance (MR) image registration aims to align similar things in different MR images spatially. Such a technology is useful in auxiliary disease diagnosis and surgical treatment. However, inconsistent intensity correspondence and large initial displacement contribute to the difficulty in registering multimodal MR volumes. A coarse-to-fine method is proposed in this study for pairwise 3D MR image rigid registration. Firstly, the proposed method extracts image feature points to form unregistered point sets and performs coarse registration based on point set registration to reduce the initial displacements of offset images effectively. Then, this method calculates a grey histogram based on voxels in the adaptive region of interest and further improves registration accuracy by maximizing mutual information of coarse-registered images. Some representative registration methods are compared on the basis of three MR image datasets to evaluate the performance of the proposed method. Experimental results show that the proposed method improved more in registration success rate and accuracy compared with conventional registration methods, especially when initial displacements are large.

Arbitrary-Shaped Building Boundary-Aware Detection With Pixel Aggregation Network

Large-scale building extraction is an essential work in the field of a remote sensing image analysis. The high-resolution image extraction methods based on deep learning have achieved state-of-the-art performance. However, most of the previous work has focused on region accuracy rather than boundary quality. Aiming at the low-accuracy problems and incomplete boundary of the building extraction method, we propose a predictive optimization architecture, BAPANet. Notably, the architecture consists of an encoder–decoder network, and residual refinement modules responsible for prediction, and refinement. The objective function optimizes the network in the form of three levels (pixel, feature map, and patch) by fusing three loss functions: binary cross-entropy, intersection over-union, and structural similarity. The five public datasets’ experimental results show that the extraction method in this article has high region accuracy, and the boundary of buildings is clear and complete.

Transformer Winding Fault Diagnosis Using Vibration Image and Deep Learning

Winding condition assessment is an essential task for operating transformers, and the vibration method provides a low-cost and non-intrusive approach. In this paper, a novel feature extraction method based on vibration analysis is proposed, which converts the vibration monitoring data with load information into a vibration image. Then, a deep learning approach based on convolutional neural network (CNN) is used to classify the images belong to different classes. In the laboratory experiment, free vibration tests are performed on an on-load winding model, which are used to verify the relationship between the natural frequency and the electromagnetic force under different clamping forces. During the field experiment, transformers are divided into three categories, including normal, degraded and anomalous, and the proposed scheme is trained and tested by using the vibration samples acquired from more than 100 operating transformers. The performance of the CNN classifier under different input sizes is investigated, which achieves 98.3% overall accuracy. Besides, the confusion matrices obtained by other methods are compared, such as artificial neural network (ANN), support vector machine (SVM) and naive Bayes classifier (NBC). The results show that the proposed scheme including the vibration image extraction method and the CNN classifier offers superior performance in winding fault diagnosis.

Improvement and Application Analysis of Lake Water Body Tracking and Recognition Algorithm

Basing on 30 years’ historical image data sources of Sanshan Lake between 1986 and 2017, this paper proposes a water body tracking and recognition algorithm according to the complex water area features of the lake and its branches to conduct the extraction of the historical water surface area and analysis on influence factors of Sanshan Lake. Comparing to conventional image extraction methods such as the region growing algorithm and the two-pass scanning method, this method shows its advantage of being fast and accurate. By analysing the water surface variation law of Sanshan Lake in 30 years and the influence factors, the area of Sanshan Lake is found to show a significant attenuation trend with the annual average attenuation rate reaches 0.98%; the water surface area weakly correlated with the precipitation change; the lake shore waterline line retreated 0.3- 10 meters in the 30 years; the pond and farmland area increased by 133%, indicating that the regulation of the lake water body is more affected by the lake shoreline dike and the Regulating capacity of gate pump.

Coastline Changes On The Coast of Cirebon Using Landsat

The coastal region is a very dynamic region because it undergoes two processes, natural and non-natural. The coastline can change due to waves, tides and human activities. Human activities are mostly carried out in coastal areas including reclamation for ports. Coastal area of Cirebon is one of the areas that experiences land addition or port reclamation which causes changes in the coastline. Images interpretation is one way to find out changes that occur in coastal areas by using Landsat 8 OLI and Landsat 5TM imagery. The image extraction method using RGB 432 can see the differences between the sea-land and how the changes has occurred in the last ten years, by overlaying both of them which results in extensive abrasion and accretion that occurs on the coast of Cirebon. The abrasion area is 107.12 Ha and the accretion is 941.17 Ha.

Estimating cement compressive strength using three-dimensional microstructure images and deep belief network

The estimation of cement compressive strength is of great significance in the quality inspections, technological designs, and engineering applications for cement. Compared to destructive methods, the nondestructive estimation approaches save the cost in the manpower and material. However, the existing nondestructive methods have the large error because the used influence factors are difficult to control and the used two-dimensional microstructure images can not reflect the specific spatial structure of the entire cement. In this paper, a novel model is proposed to estimate the cement compressive strength using three-dimensional microstructure images and deep belief network. To reduce the computation consumption induced by three-dimensional images with abundant information, this method extracts image features that reflect the cement hydration state to estimate cement compressive strength. Deep belief network is applied to build the estimation model. Its unique training pattern and flexibility of parameters improve the ability to learn nonlinear relationships between microstructure images and cement compressive strength. Furthermore, the training processes are accelerated on the graphics processing units. The experimental results prove that the proposed method estimates cement compressive strength nondestructively and improves the efficiency.