Human-human interaction recognition is crucial in computer vision fields like surveillance,human-computer interaction,and social robotics.It enhances systems’ability to interpret and respond to human behavior precise...Human-human interaction recognition is crucial in computer vision fields like surveillance,human-computer interaction,and social robotics.It enhances systems’ability to interpret and respond to human behavior precisely.This research focuses on recognizing human interaction behaviors using a static image,which is challenging due to the complexity of diverse actions.The overall purpose of this study is to develop a robust and accurate system for human interaction recognition.This research presents a novel image-based human interaction recognition method using a Hidden Markov Model(HMM).The technique employs hue,saturation,and intensity(HSI)color transformation to enhance colors in video frames,making them more vibrant and visually appealing,especially in low-contrast or washed-out scenes.Gaussian filters reduce noise and smooth imperfections followed by silhouette extraction using a statistical method.Feature extraction uses the features from Accelerated Segment Test(FAST),Oriented FAST,and Rotated BRIEF(ORB)techniques.The application of Quadratic Discriminant Analysis(QDA)for feature fusion and discrimination enables high-dimensional data to be effectively analyzed,thus further enhancing the classification process.It ensures that the final features loaded into the HMM classifier accurately represent the relevant human activities.The impressive accuracy rates of 93%and 94.6%achieved in the BIT-Interaction and UT-Interaction datasets respectively,highlight the success and reliability of the proposed technique.The proposed approach addresses challenges in various domains by focusing on frame improvement,silhouette and feature extraction,feature fusion,and HMM classification.This enhances data quality,accuracy,adaptability,reliability,and reduction of errors.展开更多
The recognition of pathological voice is considered a difficult task for speech analysis.Moreover,otolaryngologists needed to rely on oral communication with patients to discover traces of voice pathologies like dysph...The recognition of pathological voice is considered a difficult task for speech analysis.Moreover,otolaryngologists needed to rely on oral communication with patients to discover traces of voice pathologies like dysphonia that are caused by voice alteration of vocal folds and their accuracy is between 60%–70%.To enhance detection accuracy and reduce processing speed of dysphonia detection,a novel approach is proposed in this paper.We have leveraged Linear Discriminant Analysis(LDA)to train multiple Machine Learning(ML)models for dysphonia detection.Several ML models are utilized like Support Vector Machine(SVM),Logistic Regression,and K-nearest neighbor(K-NN)to predict the voice pathologies based on features like Mel-Frequency Cepstral Coefficients(MFCC),Fundamental Frequency(F0),Shimmer(%),Jitter(%),and Harmonic to Noise Ratio(HNR).The experiments were performed using Saarbrucken Voice Data-base(SVD)and a privately collected dataset.The K-fold cross-validation approach was incorporated to increase the robustness and stability of the ML models.According to the experimental results,our proposed approach has a 70%increase in processing speed over Principal Component Analysis(PCA)and performs remarkably well with a recognition accuracy of 95.24%on the SVD dataset surpassing the previous best accuracy of 82.37%.In the case of the private dataset,our proposed method achieved an accuracy rate of 93.37%.It can be an effective non-invasive method to detect dysphonia.展开更多
GIS and remote sensing were utilized for prioritizing the W. Mujib catchment. Fifty three fourth-order sub-watersheds were prioritized based on morphometric analysis of linear and shape parameters. ASTER DEM (v.2), to...GIS and remote sensing were utilized for prioritizing the W. Mujib catchment. Fifty three fourth-order sub-watersheds were prioritized based on morphometric analysis of linear and shape parameters. ASTER DEM (v.2), topographical maps, and Arc GIS (10.1) software, have been employed to delineate the 53 sub-basins, to extract the drainage networks, and to compute the required basic, linear, and shape parameters, and to compile the necessary thematic maps such as elevation and slope categories. The land use/land cover map was generated using ERDAS Imagine (2015), LANDSAT 8 image, and supervised classification (Maximum Likelihood Method). Soil map was digitized using the Arc GIS tool. Each sub-basin is prioritized by assigning ranks based on the calculated compound parameter (Cp). The final score for each sub-basin is ascribed as per erosion threat. The 53 sub-watersheds were grouped into four categories of priority: very high (15 sub-basins, 28.3% of the total), high (17 sub-basins, 32% of the total), moderate (16 sub-basins, 30.2% of the total), and low (5 sub-basins, 9.5% of the total). Sub-basins categorized as very high and high priority (60.3% of the total) are subjected to high erosion risk, thus, creating an urgent need for applying soil and water conservation measures. The validity of the prioritized four groups was tested statistically by means of Discriminant Analysis (DA), and a significant difference was found between the four priority classes. A relatively complete separation exists between the recognized priority classes;thus, they are statistically valid, distinct, and different from each other. The present results intend to help decision makers pay sufficient attention to soil and water conservation programs, and to encourage tree plantation over the government-owned sloping land. Such procedures are essential in order to minimize soil erosion loss, and to increase soil moisture on farms, thus, reducing the impact of recurrent droughts and the possibility of flooding downstream.展开更多
Air quality is a critical concern for public health and environmental regulation. The Air Quality Index (AQI), a widely adopted index by the US Environmental Protection Agency (EPA), serves as a crucial metric for rep...Air quality is a critical concern for public health and environmental regulation. The Air Quality Index (AQI), a widely adopted index by the US Environmental Protection Agency (EPA), serves as a crucial metric for reporting site-specific air pollution levels. Accurately predicting air quality, as measured by the AQI, is essential for effective air pollution management. In this study, we aim to identify the most reliable regression model among linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), logistic regression, and K-nearest neighbors (KNN). We conducted four different regression analyses using a machine learning approach to determine the model with the best performance. By employing the confusion matrix and error percentages, we selected the best-performing model, which yielded prediction error rates of 22%, 23%, 20%, and 27%, respectively, for LDA, QDA, logistic regression, and KNN models. The logistic regression model outperformed the other three statistical models in predicting AQI. Understanding these models' performance can help address an existing gap in air quality research and contribute to the integration of regression techniques in AQI studies, ultimately benefiting stakeholders like environmental regulators, healthcare professionals, urban planners, and researchers.展开更多
Linear discriminant analysis and kernel vector quantization are integrated into vector quantization based speech recognition system for improving the recognition accuracy of Mandarin digits. These techniques increase ...Linear discriminant analysis and kernel vector quantization are integrated into vector quantization based speech recognition system for improving the recognition accuracy of Mandarin digits. These techniques increase the class separability and optimize the clustering procedure. Speaker-dependent (SD) and speaker-independent (SI) experiments are performed to evaluate the performance of the proposed method. The experiment results show that the proposed method is capable of reaching the word error rate of 3.76% in SD case and 6.60 % in SI case. Such a system can be suitable for being embedded in personal digital assistant(PDA), mobile phone and so on to perform voice controlling such as digit dialing, calculating, etc.展开更多
Quantitatively determining the sources of dune sand is one of the problems necessarily and urgently to be solved in aeolian landforms and desertification research. Based on the granulometric data of sand materials fro...Quantitatively determining the sources of dune sand is one of the problems necessarily and urgently to be solved in aeolian landforms and desertification research. Based on the granulometric data of sand materials from the Hulun Buir Sandy Land, the paper employs the stepwise discriminant analysis technique (SDA) for two groups to select the principal factors determining the differences between surface loose sediments. The extent of similarity between two statistical populations can be described quantitatively by three factors such as the number of principal variables, Mahalanobis distance D 2 and confidence level 琢for F-test. Results reveal that: 1) Aeolian dune sand in the region mainly derives from Hailar Formation (Q 3 ), while fluvial sand and palaeosol also supply partially source sand for dunes; and 2) in the vicinity of Cuogang Town and west of the broad valley of the lower reaches of Hailar River, fluvial sand can naturally become principal supplier for dune sand.展开更多
In gene prediction, the Fisher discriminant analysis (FDA) is used to separate protein coding region (exon) from non-coding regions (intron). Usually, the positive data set and the negative data set are of the same si...In gene prediction, the Fisher discriminant analysis (FDA) is used to separate protein coding region (exon) from non-coding regions (intron). Usually, the positive data set and the negative data set are of the same size if the number of the data is big enough. But for some situations the data are not sufficient or not equal, the threshold used in FDA may have important influence on prediction results. This paper presents a study on the selection of the threshold. The eigen value of each exon/intron sequence is computed using the Z-curve method with 69 variables. The experiments results suggest that the size and the standard deviation of the data sets and the threshold are the three key elements to be taken into consideration to improve the prediction results.展开更多
Clinical indications for herbal drugs very often only rely on traditional knowledge. Single plant-derived preparations are used for many purposes and cannot be classified to belong to a single category like calming or...Clinical indications for herbal drugs very often only rely on traditional knowledge. Single plant-derived preparations are used for many purposes and cannot be classified to belong to a single category like calming or stimulating drugs. With respect to the brain a unique possibility exists to analyze drug effects by recording the EEG. It is common knowledge that many drugs change the frequency content of electric brain activity. Quantitative analysis of the EEG by Fast Fourier Transformation reveals parameters like spectral power, which can be processed further (CATEEM®). Source density was determined from 17 channels of the quantitative EEG from 10 clinical studies recorded in a relaxed state with open eyes. Linear discriminant analysis was used to differentiate the effects of Placebo (circadian rhythm) from CNS-active herbal drugs in comparison to Valium®. Calmvalera®, L-Theanine, Lasea®, Neurapas®, Neuravena®, Neurexan®, Nutrifin Relax®, Pascoflair®(herbal calming drugs) as well as memoLoges®, Zembrin®(herbal stimulating drugs) induced different changes of the frequency content of brain electric activity. Discriminant analysis revealed that Nutrifin Relax®, Pascoflair®and Suntheanine®could not be separated well from each other indicating a similar mechanism of action. The effect of Valium®was projected at a very isolated position far away from the herbal preparations indicating a totally different mechanism of action. Zembrin®and memoLoges®grouped together with respect to the first three discriminant functions, but were different with respect to the 4th to 6th discriminant function. Lasea®as anxiolytic drug and Neurapas®as antidepressive drug were projected at isolated positions indicating their different clinical indications. The results indicate that discriminant analysis of human quantitative EEG data allows for unique pharmacological description of individual effect profiles of herbal drugs.展开更多
The bug tracking system is well known as the project support tool of open source software. There are many categorical data sets recorded on the bug tracking system. In the past, many reliability assessment methods hav...The bug tracking system is well known as the project support tool of open source software. There are many categorical data sets recorded on the bug tracking system. In the past, many reliability assessment methods have been proposed in the research area of software reliability. Also, there are several software project analyses based on the software effort data such as the earned value management. In particular, the software reliability growth models can </span><span style="font-family:Verdana;">apply to the system testing phase of software development. On the other</span><span style="font-family:Verdana;"> hand, the software effort analysis can apply to all development phase, because the fault data is only recorded on the testing phase. We focus on the big fault data and effort data of open source software. Then, it is difficult to assess by using the typical statistical assessment method, because the data recorded on the bug tracking system is large scale. Also, we discuss the jump diffusion process model based on the estimation method of jump parameters by using the discriminant analysis. Moreover, we analyze actual big fault data to show numerical examples of software effort assessment considering many categorical data set.展开更多
We revisit a comparison of two discriminant analysis procedures, namely the linear combination classifier of Chung and Han (2000) and the maximum likelihood estimation substitution classifier for the problem of classi...We revisit a comparison of two discriminant analysis procedures, namely the linear combination classifier of Chung and Han (2000) and the maximum likelihood estimation substitution classifier for the problem of classifying unlabeled multivariate normal observations with equal covariance matrices into one of two classes. Both classes have matching block monotone missing training data. Here, we demonstrate that for intra-class covariance structures with at least small correlation among the variables with missing data and the variables without block missing data, the maximum likelihood estimation substitution classifier outperforms the Chung and Han (2000) classifier regardless of the percent of missing observations. Specifically, we examine the differences in the estimated expected error rates for these classifiers using a Monte Carlo simulation, and we compare the two classifiers using two real data sets with monotone missing data via parametric bootstrap simulations. Our results contradict the conclusions of Chung and Han (2000) that their linear combination classifier is superior to the MLE classifier for block monotone missing multivariate normal data.展开更多
Most face recognition techniques have been successful in dealing with high-resolution(HR) frontal face images. However, real-world face recognition systems are often confronted with the low-resolution(LR) face images ...Most face recognition techniques have been successful in dealing with high-resolution(HR) frontal face images. However, real-world face recognition systems are often confronted with the low-resolution(LR) face images with pose and illumination variations. This is a very challenging issue, especially under the constraint of using only a single gallery image per person.To address the problem, we propose a novel approach called coupled kernel-based enhanced discriminant analysis(CKEDA).CKEDA aims to simultaneously project the features from LR non-frontal probe images and HR frontal gallery ones into a common space where discrimination property is maximized.There are four advantages of the proposed approach: 1) by using the appropriate kernel function, the data becomes linearly separable, which is beneficial for recognition; 2) inspired by linear discriminant analysis(LDA), we integrate multiple discriminant factors into our objective function to enhance the discrimination property; 3) we use the gallery extended trick to improve the recognition performance for a single gallery image per person problem; 4) our approach can address the problem of matching LR non-frontal probe images with HR frontal gallery images,which is difficult for most existing face recognition techniques.Experimental evaluation on the multi-PIE dataset signifies highly competitive performance of our algorithm.展开更多
基金funding this work under the Research Group Funding Program Grant Code(NU/RG/SERC/12/6)supported via funding from Prince Satam bin Abdulaziz University Project Number(PSAU/2023/R/1444)+1 种基金Princess Nourah bint Abdulrahman University Researchers Supporting Project Number(PNURSP2023R348)Princess Nourah bint Abdulrahman University,Riyadh,Saudi Arabia,and this work was also supported by the Ministry of Science and ICT(MSIT),South Korea,through the ICT Creative Consilience Program supervised by the Institute for Information and Communications Technology Planning and Evaluation(IITP)under Grant IITP-2023-2020-0-01821.
文摘Human-human interaction recognition is crucial in computer vision fields like surveillance,human-computer interaction,and social robotics.It enhances systems’ability to interpret and respond to human behavior precisely.This research focuses on recognizing human interaction behaviors using a static image,which is challenging due to the complexity of diverse actions.The overall purpose of this study is to develop a robust and accurate system for human interaction recognition.This research presents a novel image-based human interaction recognition method using a Hidden Markov Model(HMM).The technique employs hue,saturation,and intensity(HSI)color transformation to enhance colors in video frames,making them more vibrant and visually appealing,especially in low-contrast or washed-out scenes.Gaussian filters reduce noise and smooth imperfections followed by silhouette extraction using a statistical method.Feature extraction uses the features from Accelerated Segment Test(FAST),Oriented FAST,and Rotated BRIEF(ORB)techniques.The application of Quadratic Discriminant Analysis(QDA)for feature fusion and discrimination enables high-dimensional data to be effectively analyzed,thus further enhancing the classification process.It ensures that the final features loaded into the HMM classifier accurately represent the relevant human activities.The impressive accuracy rates of 93%and 94.6%achieved in the BIT-Interaction and UT-Interaction datasets respectively,highlight the success and reliability of the proposed technique.The proposed approach addresses challenges in various domains by focusing on frame improvement,silhouette and feature extraction,feature fusion,and HMM classification.This enhances data quality,accuracy,adaptability,reliability,and reduction of errors.
文摘The recognition of pathological voice is considered a difficult task for speech analysis.Moreover,otolaryngologists needed to rely on oral communication with patients to discover traces of voice pathologies like dysphonia that are caused by voice alteration of vocal folds and their accuracy is between 60%–70%.To enhance detection accuracy and reduce processing speed of dysphonia detection,a novel approach is proposed in this paper.We have leveraged Linear Discriminant Analysis(LDA)to train multiple Machine Learning(ML)models for dysphonia detection.Several ML models are utilized like Support Vector Machine(SVM),Logistic Regression,and K-nearest neighbor(K-NN)to predict the voice pathologies based on features like Mel-Frequency Cepstral Coefficients(MFCC),Fundamental Frequency(F0),Shimmer(%),Jitter(%),and Harmonic to Noise Ratio(HNR).The experiments were performed using Saarbrucken Voice Data-base(SVD)and a privately collected dataset.The K-fold cross-validation approach was incorporated to increase the robustness and stability of the ML models.According to the experimental results,our proposed approach has a 70%increase in processing speed over Principal Component Analysis(PCA)and performs remarkably well with a recognition accuracy of 95.24%on the SVD dataset surpassing the previous best accuracy of 82.37%.In the case of the private dataset,our proposed method achieved an accuracy rate of 93.37%.It can be an effective non-invasive method to detect dysphonia.
文摘GIS and remote sensing were utilized for prioritizing the W. Mujib catchment. Fifty three fourth-order sub-watersheds were prioritized based on morphometric analysis of linear and shape parameters. ASTER DEM (v.2), topographical maps, and Arc GIS (10.1) software, have been employed to delineate the 53 sub-basins, to extract the drainage networks, and to compute the required basic, linear, and shape parameters, and to compile the necessary thematic maps such as elevation and slope categories. The land use/land cover map was generated using ERDAS Imagine (2015), LANDSAT 8 image, and supervised classification (Maximum Likelihood Method). Soil map was digitized using the Arc GIS tool. Each sub-basin is prioritized by assigning ranks based on the calculated compound parameter (Cp). The final score for each sub-basin is ascribed as per erosion threat. The 53 sub-watersheds were grouped into four categories of priority: very high (15 sub-basins, 28.3% of the total), high (17 sub-basins, 32% of the total), moderate (16 sub-basins, 30.2% of the total), and low (5 sub-basins, 9.5% of the total). Sub-basins categorized as very high and high priority (60.3% of the total) are subjected to high erosion risk, thus, creating an urgent need for applying soil and water conservation measures. The validity of the prioritized four groups was tested statistically by means of Discriminant Analysis (DA), and a significant difference was found between the four priority classes. A relatively complete separation exists between the recognized priority classes;thus, they are statistically valid, distinct, and different from each other. The present results intend to help decision makers pay sufficient attention to soil and water conservation programs, and to encourage tree plantation over the government-owned sloping land. Such procedures are essential in order to minimize soil erosion loss, and to increase soil moisture on farms, thus, reducing the impact of recurrent droughts and the possibility of flooding downstream.
文摘Air quality is a critical concern for public health and environmental regulation. The Air Quality Index (AQI), a widely adopted index by the US Environmental Protection Agency (EPA), serves as a crucial metric for reporting site-specific air pollution levels. Accurately predicting air quality, as measured by the AQI, is essential for effective air pollution management. In this study, we aim to identify the most reliable regression model among linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), logistic regression, and K-nearest neighbors (KNN). We conducted four different regression analyses using a machine learning approach to determine the model with the best performance. By employing the confusion matrix and error percentages, we selected the best-performing model, which yielded prediction error rates of 22%, 23%, 20%, and 27%, respectively, for LDA, QDA, logistic regression, and KNN models. The logistic regression model outperformed the other three statistical models in predicting AQI. Understanding these models' performance can help address an existing gap in air quality research and contribute to the integration of regression techniques in AQI studies, ultimately benefiting stakeholders like environmental regulators, healthcare professionals, urban planners, and researchers.
文摘Linear discriminant analysis and kernel vector quantization are integrated into vector quantization based speech recognition system for improving the recognition accuracy of Mandarin digits. These techniques increase the class separability and optimize the clustering procedure. Speaker-dependent (SD) and speaker-independent (SI) experiments are performed to evaluate the performance of the proposed method. The experiment results show that the proposed method is capable of reaching the word error rate of 3.76% in SD case and 6.60 % in SI case. Such a system can be suitable for being embedded in personal digital assistant(PDA), mobile phone and so on to perform voice controlling such as digit dialing, calculating, etc.
文摘Quantitatively determining the sources of dune sand is one of the problems necessarily and urgently to be solved in aeolian landforms and desertification research. Based on the granulometric data of sand materials from the Hulun Buir Sandy Land, the paper employs the stepwise discriminant analysis technique (SDA) for two groups to select the principal factors determining the differences between surface loose sediments. The extent of similarity between two statistical populations can be described quantitatively by three factors such as the number of principal variables, Mahalanobis distance D 2 and confidence level 琢for F-test. Results reveal that: 1) Aeolian dune sand in the region mainly derives from Hailar Formation (Q 3 ), while fluvial sand and palaeosol also supply partially source sand for dunes; and 2) in the vicinity of Cuogang Town and west of the broad valley of the lower reaches of Hailar River, fluvial sand can naturally become principal supplier for dune sand.
文摘In gene prediction, the Fisher discriminant analysis (FDA) is used to separate protein coding region (exon) from non-coding regions (intron). Usually, the positive data set and the negative data set are of the same size if the number of the data is big enough. But for some situations the data are not sufficient or not equal, the threshold used in FDA may have important influence on prediction results. This paper presents a study on the selection of the threshold. The eigen value of each exon/intron sequence is computed using the Z-curve method with 69 variables. The experiments results suggest that the size and the standard deviation of the data sets and the threshold are the three key elements to be taken into consideration to improve the prediction results.
文摘Clinical indications for herbal drugs very often only rely on traditional knowledge. Single plant-derived preparations are used for many purposes and cannot be classified to belong to a single category like calming or stimulating drugs. With respect to the brain a unique possibility exists to analyze drug effects by recording the EEG. It is common knowledge that many drugs change the frequency content of electric brain activity. Quantitative analysis of the EEG by Fast Fourier Transformation reveals parameters like spectral power, which can be processed further (CATEEM®). Source density was determined from 17 channels of the quantitative EEG from 10 clinical studies recorded in a relaxed state with open eyes. Linear discriminant analysis was used to differentiate the effects of Placebo (circadian rhythm) from CNS-active herbal drugs in comparison to Valium®. Calmvalera®, L-Theanine, Lasea®, Neurapas®, Neuravena®, Neurexan®, Nutrifin Relax®, Pascoflair®(herbal calming drugs) as well as memoLoges®, Zembrin®(herbal stimulating drugs) induced different changes of the frequency content of brain electric activity. Discriminant analysis revealed that Nutrifin Relax®, Pascoflair®and Suntheanine®could not be separated well from each other indicating a similar mechanism of action. The effect of Valium®was projected at a very isolated position far away from the herbal preparations indicating a totally different mechanism of action. Zembrin®and memoLoges®grouped together with respect to the first three discriminant functions, but were different with respect to the 4th to 6th discriminant function. Lasea®as anxiolytic drug and Neurapas®as antidepressive drug were projected at isolated positions indicating their different clinical indications. The results indicate that discriminant analysis of human quantitative EEG data allows for unique pharmacological description of individual effect profiles of herbal drugs.
文摘The bug tracking system is well known as the project support tool of open source software. There are many categorical data sets recorded on the bug tracking system. In the past, many reliability assessment methods have been proposed in the research area of software reliability. Also, there are several software project analyses based on the software effort data such as the earned value management. In particular, the software reliability growth models can </span><span style="font-family:Verdana;">apply to the system testing phase of software development. On the other</span><span style="font-family:Verdana;"> hand, the software effort analysis can apply to all development phase, because the fault data is only recorded on the testing phase. We focus on the big fault data and effort data of open source software. Then, it is difficult to assess by using the typical statistical assessment method, because the data recorded on the bug tracking system is large scale. Also, we discuss the jump diffusion process model based on the estimation method of jump parameters by using the discriminant analysis. Moreover, we analyze actual big fault data to show numerical examples of software effort assessment considering many categorical data set.
文摘We revisit a comparison of two discriminant analysis procedures, namely the linear combination classifier of Chung and Han (2000) and the maximum likelihood estimation substitution classifier for the problem of classifying unlabeled multivariate normal observations with equal covariance matrices into one of two classes. Both classes have matching block monotone missing training data. Here, we demonstrate that for intra-class covariance structures with at least small correlation among the variables with missing data and the variables without block missing data, the maximum likelihood estimation substitution classifier outperforms the Chung and Han (2000) classifier regardless of the percent of missing observations. Specifically, we examine the differences in the estimated expected error rates for these classifiers using a Monte Carlo simulation, and we compare the two classifiers using two real data sets with monotone missing data via parametric bootstrap simulations. Our results contradict the conclusions of Chung and Han (2000) that their linear combination classifier is superior to the MLE classifier for block monotone missing multivariate normal data.
基金supported by National Natural Science Foundation of China(60802069,61273270)the Fundamental Research Funds for the Central Universities of China+1 种基金Natural Science Foundation of Guangdong Province(2014A030313173)Science and Technology Program of Guangzhou(2014Y2-00165,2014J4100114,2014J4100095)
文摘Most face recognition techniques have been successful in dealing with high-resolution(HR) frontal face images. However, real-world face recognition systems are often confronted with the low-resolution(LR) face images with pose and illumination variations. This is a very challenging issue, especially under the constraint of using only a single gallery image per person.To address the problem, we propose a novel approach called coupled kernel-based enhanced discriminant analysis(CKEDA).CKEDA aims to simultaneously project the features from LR non-frontal probe images and HR frontal gallery ones into a common space where discrimination property is maximized.There are four advantages of the proposed approach: 1) by using the appropriate kernel function, the data becomes linearly separable, which is beneficial for recognition; 2) inspired by linear discriminant analysis(LDA), we integrate multiple discriminant factors into our objective function to enhance the discrimination property; 3) we use the gallery extended trick to improve the recognition performance for a single gallery image per person problem; 4) our approach can address the problem of matching LR non-frontal probe images with HR frontal gallery images,which is difficult for most existing face recognition techniques.Experimental evaluation on the multi-PIE dataset signifies highly competitive performance of our algorithm.