- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
Nondestructive Detection of Postharvest Quality of Cherry Tomatoes Using a Portable NIR Spectrometer and Chemometric Algorithms
摘要： The aim of this study was to assess the applicability of a portable NIR spectroscopy system and chemometric algorithms in intelligently detecting postharvest quality of cherry tomatoes. The postharvest quality of cherry tomatoes was evaluated in terms of firmness, soluble solids content (SSC), and pH, and a portable NIR spectrometer (950–1650 nm) was used to obtain the spectra of cherry tomatoes. Partial least square (PLS), support vector machine (SVM), and extreme learning machine (ELM) were applied to predict the postharvest quality of cherry tomatoes from their spectra. The effects of different preprocessing techniques, including Savitzky–Golay (S-G), multiplicative scattering correction (MSC), and standard normal variate (SNV) on prediction performance were also evaluated. Firmness, SSC and pH values of cherry tomatoes decreased during storage period, based on which the tomato samples could be classified into two distinct clusters. Similarly, cherry tomatoes with different storage time could also be separated by the NIR spectroscopic characteristics. The best prediction accuracy was obtained from ELM algorithms using the raw spectra with Rp2, RMSEP, and RPD values of 0.9666, 0.3141 N, and 5.6118 for firmness; 0.9179, 0.1485%, and 3.6249 for SSC; and 0.8519, 0.0164, and 2.7407 for pH, respectively. Excellent predictions for firmness and SSC (RPD value greater than 3.0), good prediction for pH (RPD value between 2.5 and 3.0) were obtained using ELM model. NIR spectroscopy is capable of intelligently detecting postharvest quality of cherry tomatoes during storage.
关键词： Partial least square,Extreme learning machine,Support vector machine,Cherry tomato,Near infrared spectroscopy
更新于2025-09-23 15:23:52
-
Computer-Assisted Diagnosis for Diabetic Retinopathy Based on Fundus Images Using Deep Convolutional Neural Network
摘要： Diabetic retinopathy (DR) is a complication of long-standing diabetes, which is hard to detect in its early stage because it only shows a few symptoms. Nowadays, the diagnosis of DR usually requires taking digital fundus images, as well as images using optical coherence tomography (OCT). Since OCT equipment is very expensive, it will benefit both the patients and the ophthalmologists if an accurate diagnosis can be made, based solely on reading digital fundus images. In the paper, we present a novel algorithm based on deep convolutional neural network (DCNN). Unlike the traditional DCNN approach, we replace the commonly used max-pooling layers with fractional max-pooling. Two of these DCNNs with a different number of layers are trained to derive more discriminative features for classification. After combining features from metadata of the image and DCNNs, we train a support vector machine (SVM) classifier to learn the underlying boundary of distributions of each class. For the experiments, we used the publicly available DR detection database provided by Kaggle. We used 34,124 training images and 1,000 validation images to build our model and tested with 53,572 testing images. The proposed DR classifier classifies the stages of DR into five categories, labeled with an integer ranging between zero and four. The experimental results show that the proposed method can achieve a recognition rate up to 86.17%, which is higher than previously reported in the literature. In addition to designing a machine learning algorithm, we also develop an app called 'Deep Retina.' Equipped with a handheld ophthalmoscope, the average person can take fundus images by themselves and obtain an immediate result, calculated by our algorithm. It is beneficial for home care, remote medical care, and self-examination.
关键词： deep convolutional neural network,mobile app,fractional max-pooling,support vector machine,diabetic retinopathy,fundus images,teaching-learning-based optimization
更新于2025-09-23 15:23:52
-
[Lecture Notes in Computational Vision and Biomechanics] Computer Aided Intervention and Diagnostics in Clinical and Medical Images Volume 31 || Retina as a Biomarker of Stroke
摘要： Stroke is one of the significant reasons of adult impairment in most of the developing nations worldwide. Various imaging modalities are used to diagnose stroke during its initial hours of occurrence. But early prediction of stroke is still a challenge in the field of biomedical research. Since retinal arterioles share similar anatomical, physiological, and embryological attributes with brain arterioles, analysis of retinal fundus images can be of great significance in stroke prognosis. This research work mainly analyzes the variations in retinal vasculature in predicting the risk of stroke. Fractal dimension, branching coefficients and angle, asymmetry factor and optimality ratio for both arteries and veins were computed from the processed input image and given to a support vector machine classifier which gives promising results.
关键词： Support vector machine,Stroke,Retinal fundus images
更新于2025-09-23 15:23:52
-
[IEEE 2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC) - Chongqing (2018.6.27-2018.6.29)] 2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC) - Instance Selection in the Projected High Dimensional Feature Space for SVM
摘要： Data classi?cation is a supervised learning task where a training set with previously known information is used to construct a classi?er. The classi?er is then used to predict the class of unforeseen test instances. It is often bene?cial to use a subset of the training set to construct the classi?er, in particular when the size of the data set is large. For example, support vector machine (SVM), one of the most effective classi?ers, only needs the support vectors to make the prediction. Therefore, all non-support vectors can be eliminated without affecting the classi?cation performance. However, it is usually unknown which instances in the training set are support vectors before the training is completed. Researchers have developed different methods to delete the potential non-support vectors while retaining the likely support vectors before the training starts. This preprocessing to the training data set is often known as instance selection. Many of the instance selection methods are based on the geometry of the training samples. Measures in the original feature space are usually used. We propose to use measures in the projected high dimensional feature space for SVM since this is where the separating hyperplanes are determined. We compare the performance with some existing methods on a few benchmark data sets. The experiments show that using measures in the projected feature space may improve the classi?cation accuracy, sometimes substantially.
关键词： data classi?cation,SVM,instance selection,support vector machine
更新于2025-09-23 15:23:52
-
Multiple deep-belief-network-based spectral-spatial classification of hyperspectral images
摘要： A deep-learning-based feature extraction has recently been proposed for HyperSpectral Images (HSI) classification. A Deep Belief Network (DBN), as part of deep learning, has been used in HSI classification for deep and abstract feature extraction. However, DBN has to simultaneously deal with hundreds of features from the HSI hyper-cube, which results into complexity and leads to limited feature abstraction and performance in the presence of limited training data. Moreover, a dimensional-reduction-based solution to this issue results in the loss of valuable spectral information, thereby affecting classification performance. To address the issue, this paper presents a Spectral-Adaptive Segmented DBN (SAS-DBN) for spectral-spatial HSI classification that exploits the deep abstract features by segmenting the original spectral bands into small sets/groups of related spectral bands and processing each group separately by using local DBNs. Furthermore, spatial features are also incorporated by first applying hyper-segmentation on the HSI. These results improved data abstraction with reduced complexity and enhanced the performance of HSI classification. Local application of DBN-based feature extraction to each group of bands reduces the computational complexity and results in better feature extraction improving classification accuracy. In general, exploiting spectral features effectively through a segmented-DBN process and spatial features through hyper-segmentation and integration of spectral and spatial features for HSI classification has a major effect on the performance of HSI classification. Experimental evaluation of the proposed technique on well-known HSI standard data sets with different contexts and resolutions establishes the efficacy of the proposed techniques, wherein the results are comparable to several recently proposed HSI classification techniques.
关键词： hyperspectral image classification,support vector machine,deep belief network,segmentation
更新于2025-09-23 15:23:52
-
Learning Dual Geometric Low-Rank Structure for Semisupervised Hyperspectral Image Classification
摘要： Most of the available graph-based semisupervised hyperspectral image classification methods adopt the cluster assumption to construct a Laplacian regularizer. However, they sometimes fail due to the existence of mixed pixels whose recorded spectra are a combination of several materials. In this paper, we propose a geometric low-rank Laplacian regularized semisupervised classifier, by exploring both the global spectral geometric structure and local spatial geometric structure of hyperspectral data. A new geometric regularized Laplacian low-rank representation (GLapLRR)-based graph is developed to evaluate spectral-spatial affinity of mixed pixels. By revealing the global low-rank and local spatial structure of images via GLapLRR, the constructed graph has the characteristics of spatial–spectral geometry description, robustness, and low sparsity, from which a more accurate classification of mixed pixels can be achieved. The proposed method is experimentally evaluated on three real hyperspectral datasets, and the results show that the proposed method outperforms its counterparts, when only a small number of labeled instances are available.
关键词： Dual geometric low-rank structure,mixed pixels,spectral-spatial affinity,hyperspectral image classification (HIC),support vector machine,semisupervised
更新于2025-09-23 15:23:52
-
[IEEE 2018 25th International Conference on Mechatronics and Machine Vision in Practice (M2VIP) - Stuttgart, Germany (2018.11.20-2018.11.22)] 2018 25th International Conference on Mechatronics and Machine Vision in Practice (M2VIP) - Vehicle and Pedestrian Recognition Using Multilayer Lidar based on Support Vector Machine
摘要： Moving-object tracking (estimating position and velocity of moving objects) is a key technology for autonomous driving systems and driving assistance systems in mobile robotics and vehicle automation domains. To predict and avoid collisions, the tracking system has to recognize objects as accurately as possible. This paper presents a method for recognizing vehicles (cars and bicyclists) and pedestrians using multilayer lidar (3D lidar). Lidar data are clustered, and eight-dimensional features are extracted from each of clustered lidar data, such as distance from the lidar, velocity, object size, number of lidar-measurement points, and distribution of reflection intensities. A multiclass support vector machine is applied to classify cars, bicyclists, and pedestrians from these features. Experiments using “The Stanford Track Collection” data set allow us to compare the proposed method with a method based on the random forest algorithm and a conventional 26-dimensional feature-based method. The comparison shows that the proposed method improves recognition accuracy and processing time over the other methods. Therefore, the proposed method can work well under low computational environments.
关键词： multiclass classification,support vector machine,low-dimensional features,multilayer lidar,object recognition
更新于2025-09-23 15:22:29
-
[IEEE 2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC) - Chongqing (2018.6.27-2018.6.29)] 2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC) - Target Detection Algorithm Based on Chamfer Distance Transform and Random Template
摘要： To complete the target detection task in the scene of a few target samples and low configuration software running environment, a target detection algorithm based on Chamfer distance transformation and random template is proposed in this paper. Firstly, construct the multi-level pyramids of the images to be searched, add random curve segments on the original template to make multiple random templates to train the SVM and use these random templates in the subsequent template matching process; Then, perform template matching from the top level, use SVM to determine whether the target position is a correct match; Finally, map the location to the next level for finer matching positioning, repeat the process until the bottom is reached. Test results show that this algorithm runs fast and has high accuracy in positioning, which make it competitive in real application.
关键词： edge extraction,support vector machine,target detection,chamfer distance transform,random template
更新于2025-09-23 15:22:29
-
[IEEE 2018 IEEE 21st International Multi-Topic Conference (INMIC) - Karachi, Pakistan (2018.11.1-2018.11.2)] 2018 IEEE 21st International Multi-Topic Conference (INMIC) - Textural and Intensity Feature Based Retinal Vessels Classification for the Identification of Hypertensive Retinopathy
摘要： Hypertensive retinopathy is a retinal disease which results as a consequence of high blood pressure. Its early detection is necessary in reducing the likelihood of permanent visual damage. The percentage of people suffering from Hypertension is high, so it is required to develop a system which automatically detects the presence of this disease. High blood pressure damages retinal vessels and due to which arteries width is reduced. This damage can be analyzed by extracting the blood vessels, classifying the segmented vessels into veins and arteries and their Arteriovenous Ratio, which is an important measure to establish whether a person is suffering from Hypertensive Retinopathy or not. This research presents a technique for automatic classification of blood vessels of retina using different classifiers and the performance of each classifier is compared on same feature set. A novel combination of features is used for classification of vessels, which is an essential step for calculation of Arteriovenous Ratio and subsequently the detection of Hypertensive Retinopathy. MATLAB has been used for this research. The results that are achieved using the proposed feature set show’s 89% accuracy.
关键词： Accuracy (ACC),Arteriovenous Ratio (AVR),Support Vector Machine (SVM),Armed Forces Institute of Ophthalmology (AFIO),Hypertensive Retinopathy (HR)
更新于2025-09-23 15:22:29
-
[ACM Press the 3rd International Conference - Seoul, Republic of Korea (2018.08.22-2018.08.24)] Proceedings of the 3rd International Conference on Biomedical Signal and Image Processing - ICBIP '18 - Automatic Detection of Cell Regions in Microscope Images Based on BFED Algorithm
摘要： Circulating tumor cells (CTC) attract attention as a biomarker that can evaluate cancer metastasis and therapeutic effects. The CTC exists in the blood of cancer patients, so pathologists analyze blood by using a fluorescence microscope. However, manual analysis by pathologists is hard-work since the number of CTC to substances contained in the blood is very few and the cell regions are often unclear depending on shooting environments. In addition, there are few studies on automatic identification of CTC. In this paper, we develop an automatic detection method of cell regions in microscope images based on bacterial foraging-based edge detection (BFED) algorithm to analyze CTC. In the first step, we detect the initial cell regions by BFED algorithm. Second, we identify whether the region is a single cell or multiple cells come in connect with other cell(s) by SVM. Third, when a cell is connected with other one, we separate the connecting cells by branch and bound algorithm and obtain the final cell regions. We applied our proposed method to 1680 microscopy images (6 cases). The experimental results demonstrate that the proposed method has a true positive rate of 93.9% and a false positive 1.29 /case.
关键词： Saliency map,Computer aided diagnosis,Support vector machine,Branch and bound algorithm,Circulating tumor cells,BFED algorithm
更新于2025-09-23 15:22:29