Geometric Image Features Research Articles

Neodymium oxide concentration state recognition model in neodymium molten salt electrolysis process based on flame image features

Accurate recognition of the concentration state of neodymium-oxide (Nd2O3) in the neodymium (Nd) molten salt electrolysis process provides crucial feedback information for online adjustment of process parameters during Nd metal batch production. This paper proposes a Nd2O3 concentration state recognition model based on flame image features. The model employs the lightweight flame segmentation algorithm (FEC-Net) to accurately segment the flame regions in the collected Nd electrolysis cell flame images. Then, using image processing techniques, geometric and color features of the flame regions are extracted. Finally, a Support Vector Machine (SVM) is used to establish a nonlinear mapping model between the flame features and Nd2O3 concentration states. This model classifies the collected flame image dataset to verify the Nd2O3 concentration state during a certain period of Nd molten salt electrolysis. Experiments demonstrate that the recognition accuracy of SVM reaches 96.09%, meeting the monitoring requirements of the Nd molten salt electrolysis process.

Multiresolution Interpretable Contourlet Graph Network for Image Classification.

Modeling contextual relationships in images as graph inference is an interesting and promising research topic. However, existing approaches only perform graph modeling of entities, ignoring the intrinsic geometric features of images. To overcome this problem, a novel multiresolution interpretable contourlet graph network (MICGNet) is proposed in this article. MICGNet delicately balances graph representation learning with the multiscale and multidirectional features of images, where contourlet is used to capture the hyperplanar directional singularities of images and multilevel sparse contourlet coefficients are encoded into graph for further graph representation learning. This process provides interpretable theoretical support for optimizing the model structure. Specifically, first, the superpixel-based region graph is constructed. Then, the region graph is applied to code the nonsubsampled contourlet transform (NSCT) coefficients of the image, which are considered as node features. Considering the statistical properties of the NSCT coefficients, we calculate the node similarity, i.e., the adjacency matrix, using Mahalanobis distance. Next, graph convolutional networks (GCNs) are employed to further learn more abstract multilevel NSCT-enhanced graph representations. Finally, the learnable graph assignment matrix is designed to get the geometric association representations, which accomplish the assignment of graph representations to grid feature maps. We conduct comparative experiments on six publicly available datasets, and the experimental analysis shows that MICGNet is significantly more effective and efficient than other algorithms of recent years.

Geometrical recognition of metallic foam microstructures using a deep learning approach

The geometric characteristics of nanostructured porous metallic foams (NMFs) impact their properties, enabling applications in electrocatalysis, energy conversion, and energy storage. The morphology of these three-dimensional (3D) structures is complex, in large part, because of their complicated formation mechanisms. This work presents a computational strategy for geometrically characterizing different NMFs designs. Three types of samples were synthesized using the dynamic hydrogen bubble template electrodeposition method. A set of microscopic confocal images of NMFs was used as training input samples. Herein, a variational autoencoder (VAE) was adjusted under a pretext task, allowing learning embedding descriptors to represent the geometry of microscopic observations. To evaluate the feasibility of using a VAE to assist in the classification of NMFs, a set of machine learning classifiers was implemented to discriminate among the groups of embedded vectors according to their classes. Furthermore, the capacity of the VAE was validated regarding the capability to separate the vectors according to their classes. Explainability mechanisms stand out geometrical features of input images that had major support during the classification task.

The use of color elements in graphic design based on convolutional neural network model

Abstract Exploring the effective application of color elements in graphic design based on a convolutional neural network model is beneficial to promote the innovative development of the graphic design. Starting from the convolutional neural network, this paper defines the quantification of geometric features of graphic images by convolutional operation and pooling operation and uses pixel and spatial distribution for color element extraction. The global color features and local color features are used for color matching perception feature quantification, and then a graphic image design feature model is constructed, and the application analysis of color elements is carried out for this model. From the primary design color, the average selection rate of this paper’s model is 41.58%, which is 13.4% and 9.6% higher than that of the DCGAN and PLS-MLP models, respectively. From the keyword color quality generation, the color design results of graphic images have significant differences (p < 0.05). This indicates that the use of CNN-based graphic image design feature models can realize the effective application of color elements and also provides a new research direction for intelligent graphic design.

Analysis and Recognition of Voluntary Facial Expression Mimicry Based on Depressed Patients.

Many clinical studies have shown that facial expression recognition and cognitive function are impaired in depressed patients. Different from spontaneous facial expression mimicry (SFEM), 164 subjects (82 in a case group and 82 in a control group) participated in our voluntary facial expression mimicry (VFEM) experiment using expressions of neutrality, anger, disgust, fear, happiness, sadness and surprise. Our research is as follows. First, we collected a large amount of subject data for VFEM. Second, we extracted the geometric features of subject facial expression images for VFEM and used Spearman correlation analysis, a random forest, and logistic regression-based recursive feature elimination (LR-RFE) to perform feature selection. The features selected revealed the difference between the case group and the control group. Third, we combined geometric features with the original images and improved the advanced deep learning facial expression recognition (FER) algorithms in different systems. We propose the E-ViT and E-ResNet based on VFEM. The accuracies and F1 scores were higher than those of the baseline models, respectively. Our research proved that it is effective to use feature selection to screen geometric features and combine them with a deep learning model for depression facial expression recognition.

Directional lifting wavelet transform for image edge analysis

In this paper, we propose a new two-dimensional directional discrete wavelet transform that can decompose an image into 12 multiscale directional edge components. The proposed transform is designed in a fully discrete setting and thus is easy to implement in actual computations. The proposed transform is viewed as a category of redundant discrete wavelet transforms implemented by fast in-place computational algorithms by a lifting scheme that has been modified to incorporate redundancy. The redundancy is limited to (N×J+1)/4, where N=12 is the directional selectivity and J is a decomposition level of the multiscale transform. Numerical experiments in edge analysis using various images demonstrate the advantages of the proposed method over some conventional standard methods. The proposed method outperforms several conventional edge detection methods in identifying both the location and orientation of edges, and well captures the directional and geometrical features of images.

Contour Primitive of Interest Extraction Network Based on Dual-Metric One-Shot Learning for Vision Measurement

Although many existing vision measurement systems have achieved high performances, they are object-specific and have limitations in flexibility. Toward intelligent vision measurement that can be conveniently reused for novel objects, this article focuses on the image geometric feature extraction with one-shot learning ability. We propose a contour primitive of interest (CPI) extraction network with dual metric (CPieNet-DM), which can obtain a designated CPI in a query image of a novel object under the guidance of only one annotated support image. First, the dual-metric learning mechanism is proposed, which not only utilizes inter-image similarity as guidance but also leverages the intra-image coherency of CPI pixels to facilitate the inference. Second, a neural network is designed to infer the CPI map based on the dual metric, which also predicts the CPI's geometric parameters. Moreover, the dual context aggregator is plugged in to provide the awareness of both images’ contexts. Third, the network training is jointly supervised by the multiple tasks of dual-metric learning, geometric parameters regression, and CPI extraction. The online hard example mining is utilized to improve the training outcome. The effectiveness of the proposed methods is validated with a series of experiments.

Rapid Three-Dimensional Reconstruction of Underwater Defective Pile Based on Two-Dimensional Images Obtained Using Mechanically Scanned Imaging Sonar

Three-dimensional (3-D) reconstruction based on sonar imaging is an intuitive and comprehensible form for describing an underwater defective pile. Most existing 3-D reconstruction methods are inefficient as they require the extraction of numerous feature points from many sonar images. A rapid 3-D reconstruction method for underwater defective piles based on two-dimensional (2-D) images obtained using mechanically scanned imaging sonar (MSIS) is proposed herein. First, the correspondences between the geometric features of MSIS images and actual defects are established based on the mapping relationship between actual object and MSIS image. Subsequently, no more than three feature points are extracted from 2-D MSIS images to recognize the contour segments. The contour segments are assembled into the entire cross section and then into the entire defective pile. The applicability of the proposed method is verified via a 3-D reconstruction for a scaled-down pile with multiple defect distributions.

Mapping Pu'er tea plantations from GF-1 images using Object-Oriented Image Analysis (OOIA) and Support Vector Machine (SVM).

Tea is the most popular drink worldwide, and China is the largest producer of tea. Therefore, tea is an important commercial crop in China, playing a significant role in domestic and foreign markets. It is necessary to make accurate and timely maps of the distribution of tea plantation areas for plantation management and decision making. In the present study, we propose a novel mapping method to map tea plantation. The town of Menghai in the Xishuangbanna Dai Autonomous Prefecture, Yunnan Province, China, was chosen as the study area, andgg GF-1 remotely sensed data from 2014-2017 were chosen as the data source. Image texture, spectral and geometrical features were integrated, while feature space was built by SEparability and THresholds algorithms (SEaTH) with decorrelation. Object-Oriented Image Analysis (OOIA) with a Support Vector Machine (SVM) algorithm was utilized to map tea plantation areas. The overall accuracy and Kappa coefficient ofh the proposed method were 93.14% and 0.81, respectively, 3.61% and 0.05, 6.99% and 0.14, 6.44% and 0.16 better than the results of CART method, Maximum likelihood method and CNN based method. The tea plantation area increased by 4,095.36 acre from 2014 to 2017, while the fastest-growing period is 2015 to 2016.

High speed melt flow monitoring and development of an image processing algorithm for laser fusion cutting

This contribution presents high-speed camera monitoring of melt pool dynamics for steel during laser fusion cutting and compares the images with recordings in aluminum. The experiments are performed by a 4 kW multimode fiber laser with an emission wavelength of 1070 nm. To visualize the thermal radiation from the process zone during the cutting process, the kerf is captured at sample rates of up to 170 000 frames per second without external illumination with a spectral response between 400 and 700 nm, allowing measurements of the melt flow dynamics from geometric image features. The dependencies of the melt flow dynamics on laser processing parameters, such as feed rate, gas pressure, and laser power, can be evaluated. The monitoring system is placed both off-axis and mounted to a conventional cutting head, with the monitoring path aligned to the processing laser for a coaxial and lateral view of the cut kerf. The measured signal characteristics of the images captured from the melt pool are examined in the visible spectral range of the emitted thermal radiation from the process zone. Moreover, a specifically developed image processing algorithm is developed that process and analyze the captured images and extract geometric information for a measurement of the melt flow.

Identification technology based on geometric features of tooth print images

Identity recognition technology is a type of technology that realizes identity verification based on certain biological characteristics. After entering the Internet era, this technology has become a popular research direction in the computer field. In this paper, the image of the tooth print is used as the biological feature to carry out the research on the identification algorithm. This paper adopts the target detection algorithm based on neural network to detect a single tooth imprint area of the target, build a target detection network. The experimental results show that the method has a good segmentation effect on the target area, and the accuracy rate is 91.66%. According to the contour features of the collected tooth print images, a set of tooth pore area ratio feature extraction methods are designed. To objectively evaluate the recognition and classification method, the support vector machine is used as the final classifier. The recognition accuracy rate is 94.09%, and the verification accuracy rate is 94.09%. The test accuracy rate is 91.46%, and the classification effect is excellent. This paper has made a lot of breakthroughs and obvious progress based on the previous research on the tooth impression model.

RETRACTED: Optimal skin cancer detection by a combined ENN and Fractional Order Coot Optimization Algorithm.

In recent years, skin cancer has been recognized as the most dangerous and common type of cancer in humans. There are different types of skin cancer that can be diagnosed early. Providing a method that facilitates the diagnosis of this cancer in the early stages is very useful and valuable. This study proposes a new algorithm to efficient diagnosis of the skin cancer by proper extracting and selection of the features from the image (gradient, texture, and geometric features) and classifying these features to diagnose the cancerous cases. The classification step of the methodology based on Elman neural network. Due to the effectiveness of the metaheuristic algorithms in optimal solving the problems in the last decade, using these kinds of optimization techniques is useful for the diagnosis purpose. Therefore, both parts of feature selection and classification stages are optimized by an improved metaheuristic algorithm, called Fractional Order Coot Optimization Algorithm (FO-COA) which is designed in this study. Final designed model is then performed to the SIIM-ISIC Melanoma dataset and its achievements have been validated with several newest approaches to show the method higher efficiency.

Radiogenomics influence on the future of prostate cancer risk stratification

In an era of powerful computing tools, radiogenomics provides a personalized, precise approach to the detection and diagnosis in patients with prostate cancer (PCa). Radiomics data are obtained through artificial intelligence (AI) and neural networks that analyze imaging, usually MRI, to assess statistical, geometrical, and textural features of images to provide quantitative data of shape, heterogeneity, and intensity of tumors. Genomics involves assessing the genomic markers that are present from tumor biopsies. In this article, we separately investigate the current landscape of radiomics and genomics within the realm of PCa and discuss the integration and validity of both into radiogenomics using the data from three papers on the topic. We also conducted a clinical trials search using the NIH’s database, where we found two relevant actively recruiting studies. Although there is more research needed to be done on radiogenomics to fully adopt it as a viable diagnosis tool, its potential by providing personalized data regarding each tumor cannot be overlooked as it may be the future of PCa risk-stratification techniques.

Triple screening point cloud registration method based on image and geometric features

Three-dimensional reconstruction of cochlear model is commonly used to measure the cochlear contour of patients with high accuracy and configure appropriate hearing aids. To improve the matching accuracy of 3D reconstruction of cochlear model, a triple screening point cloud registration method based on image and geometric features is proposed in this paper. Firstly, extracting the image feature points by SIFT algorithm, then eliminating the mismatched point pairs according to the RANSAC algorithm. Secondly, the average value of the normal feature is introduced to select the feature points with obvious geometric feature region. Thirdly, based on the point cloud fast point feature histogram (FPFH), the matching feature point pairs with image features and geometric features are selected to complete the coarse registration. Finally, iterative closest point (ICP) algorithm is used to achieve fine registration. The proposed algorithm is more comprehensive and effective for finding feature points then ensures the accuracy of registration results. In addition, results of the simulation show that the proposed algorithm reduces both the matching error and the total number of iterations compared with some existing algorithms.

Active alignment of space astronomical telescopes by matching arbitrary multi-field stellar image features.

Space-based optical astronomical telescopes are susceptible to mirror misalignments due to space disturbance in mechanics and temperature. Therefore, it is of great importance to actively align the telescope in orbit to continuously maintain imaging quality. Traditional active alignment methods usually need additional delicate wavefront sensors and complicated operations (such as instrument calibration and pointing adjustment). This paper proposes a novel active alignment approach by matching the geometrical features of several stellar images at arbitrary multiple field positions. Based on nodal aberration theory and Fourier optics, the relationship between stellar image intensity distribution and misalignments of the system can be modeled for an arbitrary field position. On this basis, an objective function is established by matching the geometrical features of the collected multi-field stellar images and modeled multi-field stellar images, and misalignments can then be solved through nonlinear optimization. Detailed simulations and a real experiment are performed to demonstrate the effectiveness and practicality of the proposed approach. This approach eliminates the need for delicate wavefront sensors and pointing adjustment, which greatly facilitates the maintainance of imaging quality.

Comparison of Support Vector Machine and K-Nearest Neighbor for Baby Foot Identification based on Image Geometric Characteristics

Biometric recognition of infant identification systems is critical in security access for identification and verification systems. However, until now, hospitals or health centres in Indonesia still use conventional biometric identification, such as stamping or inking on the soles of babies' feet affixed to paper and are very vulnerable to the risk of damage or loss of data. To resolve this problem, computer vision technology can accurately identify the baby's feet' soles with the final result in the form of digital data. This study compares the classification method of baby feet using the SVM (Support Vector Machine) algorithm with the K-Nearest Neighbor algorithm. The baby's feet understudy image was taken using a cellphone camera with sample data of 3 months old babies. Comparing the SVM and KNN classification methods obtained high accuracy, precision and recall values, namely 98.80% accuracy, 89.51% precision and 88.00% recall. (for the SVM Gaussian kernel classification), with an accuracy of 99.08%, 92.65% precision and 90.75% recall (for the KNN Ecluidean Distance classification), it can be concluded that the KNN classification method using Euclidean distance is the best for applied in the baby palm identification system using the geometric image feature.

Statistical image watermarking using local RHFMs magnitudes and beta exponential distribution

The imperceptibility, robustness and data payload are widely considered as the three main properties vital for any image watermarking systems. They are complimentary to each other and hence challenging to attain the right balance between them. The statistical model-based multiplicative watermarking is an effective way to achieve the tradeoff among imperceptibility, robustness and data payload. Radial harmonic Fourier moments (RHFMs) is a strong tool in image processing, which has many advantages, such as lower noise sensitivity, powerful image description ability and geometric invariance feature. In this paper, we propose a new statistical image watermarking scheme using local RHFMs magnitudes and Beta exponential distribution model. Our image watermarking scheme consists of two parts, namely, embedding and extraction. In the embedding process, we divide the host image into no-overlapping blocks and compute the local RHFMs of image blocks, then insert the watermark signal into the robust local RHFMs magnitudes through multiplicative approach. In the extraction phase, robust local RHFMs magnitudes are firstly modeled by employing the Beta exponential distribution, where the statistical properties of local RHFMs magnitudes are captured accurately. Then the modified maximum likelihood parameter estimation (MMLE) approach is introduced to estimate the statistical parameters of Beta exponential distribution model. And finally an image watermark decoder for multiplicative watermarking is developed using Beta exponential distribution and maximum likelihood decision criterion. Experimental results on some test images and comparison with well-known existing methods demonstrate the efficacy and superiority of the proposed statistical image watermarking.

Recognition of micro-objects with adaptive models of image processing in a parallel computing environment

Scientific and methodological foundations for optimizing the processing of micro-objects, in particular, pollen grains, have been developed on the basis of models and methods of preliminary information processing with mechanisms for filtering, identifying, and using textural, specific characteristics, and geometric features of images. The efficiency of image filtering mechanisms is investigated based on the use of statistical control rules, adaptive two-threshold control, trend functions, and control of the contour description error, traditional detectors of random noise, and interference. The mechanisms of image identification using biquadratic, orthogonal algebraic polynomials, 4th order splines are proposed. A technique for optimizing image identification by finding a function by its integrals in a family of straight lines based on the Mellin and Fourier transforms has been developed. Combined models with mechanisms for smoothing reference points, segmentation of contours, filtering with detection of changes in dynamics, optimization by the conjugate gradient method, time relaxation have been built. A software complex for the recognition and classification of images of pollen grains in C ++ has been developed, which includes testing identification algorithms based on the Daubechies spline function, orthogonal algebraic polynomials of 4, 8 orders in the parallel computing environment “CUDA”.

SMSIR: Spherical Measure Based Spherical Image Representation.

This paper presents a spherical measure based spherical image representation(SMSIR) and sphere-based resampling methods for generating our representation. On this basis, a spherical wavelet transform is also proposed. We first propose a formal recursive definition of the spherical triangle elements of SMSIR and a dyadic index scheme. The index scheme, which supports global random access and needs not to be pre-computed and stored, can efficiently index the elements of SMSIR like planar images. Two resampling methods to generate SMSIR from the most commonly used ERP(Equirectangular Projection) representation are presented. Notably, the spherical measure based resampling, which exploits the mapping between the spherical and the parameter domain, achieves higher computational efficiency than the spherical RBF(Radial Basis Function) based resampling. Finally, we design high-pass and low-pass filters with lifting schemes based on the dyadic index to further verify the efficiency of our index and deal with the spherical isotropy. It provides novel Multi-Resolution Analysis(MRA) for spherical images. Experiments on continuous synthetic spherical images indicate that our representation can recover the original image signals with higher accuracy than the ERP and CMP(Cubemap) representations at the same sampling rate. Besides, the resampling experiments on natural spherical images show that our resampling methods outperform the bilinear and bicubic interpolations concerning the subjective and objective quality. Particularly, as high as 2dB gain in terms of S-PSNR is achieved. Experiments also show that our spherical image transform can capture more geometric features of spherical images than traditional wavelet transform.

Application of computer image segmentation algorithm based on visual features

In order to further satisfy the needs of visual perception, a computer image segmentation algorithm based on visual characteristics is proposed. The computer image geometric features are linearly segmented to obtain multiple sub-panels and various geometric features of each sub-panel are extracted. The geometric features are input as low-level features into the deep neural network model to learn to generate high-level features. Finally, based on the high-level features, the clustering center is obtained by Gaussian mixture model, and the final segmentation result is obtained by using graph cut. The results of experiments on Princeton standard data set and COSEG dataset show that the rand index RI value is the best value for each 3D model, which shows that the proposed method is better than the traditional segmentation method. It has good consistent segmentation results. The research showed that using a variety of geometric features compared to a single geometric feature, the obtained features have a more comprehensive geometric meaning, which can effectively make image segmentation meet the visual characteristics requirements.