Using digital image classification methods and hyperspectral imaging to detect leaf blast in rice

2013 ◽  
Author(s):  
Lulu WU ◽  
Long QI ◽  
Xu MA ◽  
Zhixiong ZHENG ◽  
Guorui CHEN ◽  
...  
Keyword(s):  
Image Classification ◽  
Hyperspectral Imaging ◽  
Digital Image ◽  
Classification Methods ◽  
Leaf Blast

scholarly journals Effective combining of color and texture descriptors for indoor-outdoor image classification

2014 ◽  
Vol 27 (3) ◽  
pp. 399-410 ◽  
Author(s):  
Stevica Cvetkovic ◽  
Sasa Nikolic ◽  
Slobodan Ilic
Keyword(s):  
Image Classification ◽  
Accurate Method ◽  
Optimal Combination ◽  
Compact Representation ◽  
Svm Classifier ◽  
Classification Methods ◽  
Empirical Tests ◽  
Public Datasets ◽  
Fast Extraction ◽  
Classification Procedures

Although many indoor-outdoor image classification methods have been proposed in the literature, most of them have omitted comparison with basic methods to justify the need for complex feature extraction and classification procedures. In this paper we propose a relatively simple but highly accurate method for indoor-outdoor image classification, based on combination of carefully engineered MPEG-7 color and texture descriptors. In order to determine the optimal combination of descriptors in terms of fast extraction, compact representation and high accuracy, we conducted comprehensive empirical tests over several color and texture descriptors. The descriptors combination was used for training and testing of a binary SVM classifier. We have shown that the proper descriptors preprocessing before SVM classification has significant impact on the final result. Comprehensive experimental evaluation shows that the proposed method outperforms several more complex indoor-outdoor image classification techniques on a couple of public datasets.

scholarly journals Extracting image features for classification by two-tier genetic programming

2020 ◽  
Author(s):  
Harith Al-Sahaf ◽  
A Song ◽  
K Neshatian ◽  
Mengjie Zhang
Keyword(s):  
Genetic Programming ◽  
Image Classification ◽  
Domain Knowledge ◽  
Extraction Process ◽  
High Accuracy ◽  
Classification Performance ◽  
Image Features ◽  
Classification Methods ◽  
Feature Based ◽  
Second Tier

Image classification is a complex but important task especially in the areas of machine vision and image analysis such as remote sensing and face recognition. One of the challenges in image classification is finding an optimal set of features for a particular task because the choice of features has direct impact on the classification performance. However the goodness of a feature is highly problem dependent and often domain knowledge is required. To address these issues we introduce a Genetic Programming (GP) based image classification method, Two-Tier GP, which directly operates on raw pixels rather than features. The first tier in a classifier is for automatically defining features based on raw image input, while the second tier makes decision. Compared to conventional feature based image classification methods, Two-Tier GP achieved better accuracies on a range of different tasks. Furthermore by using the features defined by the first tier of these Two-Tier GP classifiers, conventional classification methods obtained higher accuracies than classifying on manually designed features. Analysis on evolved Two-Tier image classifiers shows that there are genuine features captured in the programs and the mechanism of achieving high accuracy can be revealed. The Two-Tier GP method has clear advantages in image classification, such as high accuracy, good interpretability and the removal of explicit feature extraction process. © 2012 IEEE.

scholarly journals AIRBORNE HYPERSPECTRAL REMOTE SENSING FOR IDENTIFICATION GRASSLAND VEGETATION

2015 ◽  
Vol XL-3/W3 ◽  
pp. 427-431 ◽  
Author(s):  
P. Burai ◽  
T. Tomor ◽  
L. Bekő ◽  
B. Deák
Keyword(s):  
Image Classification ◽  
Learning Algorithm ◽  
Training Sample ◽  
Hyperspectral Data ◽  
Training Dataset ◽  
Classification Methods ◽  
Grassland Vegetation ◽  
Training Samples ◽  
Almost All ◽  
Noise Fraction

In our study we classified grassland vegetation types of an alkali landscape (Eastern Hungary), using different image classification methods for hyperspectral data. Our aim was to test the applicability of hyperspectral data in this complex system using various image classification methods. To reach the highest classification accuracy, we compared the performance of traditional image classifiers, machine learning algorithm, feature extraction (MNF-transformation) and various sizes of training dataset. Hyperspectral images were acquired by an AISA EAGLE II hyperspectral sensor of 128 contiguous bands (400–1000 nm), a spectral sampling of 5 nm bandwidth and a ground pixel size of 1 m. We used twenty vegetation classes which were compiled based on the characteristic dominant species, canopy height, and total vegetation cover. Image classification was applied to the original and MNF (minimum noise fraction) transformed dataset using various training sample sizes between 10 and 30 pixels. In the case of the original bands, both SVM and RF classifiers provided high accuracy for almost all classes irrespectively of the number of the training pixels. We found that SVM and RF produced the best accuracy with the first nine MNF transformed bands. Our results suggest that in complex open landscapes, application of SVM can be a feasible solution, as this method provides higher accuracies compared to RF and MLC. SVM was not sensitive for the size of the training samples, which makes it an adequate tool for cases when the available number of training pixels are limited for some classes.

scholarly journals Comparison of Three Image Classification Methods in Urban Environment

2015 ◽  
Vol 03 (02) ◽  
pp. 54-59 ◽  
Author(s):  
Malgorzata Verőné Wojtaszek ◽  
Valéria Balázsik ◽  
Tamás Jancsó ◽  
Margit Horoszné Gulyás ◽  
Qingyan Meng
Keyword(s):  
Image Classification ◽  
Urban Environment ◽  
Classification Methods

Utilizing geometric attributes of spatial information to improve digital image classification

1998 ◽  
Vol 16 (4) ◽  
pp. 233-253 ◽  
Author(s):  
Sunil Narumalani ◽  
Yingchun Zhou ◽  
Dennis E. Jelinski
Keyword(s):  
Image Classification ◽  
Digital Image ◽  
Spatial Information

scholarly journals 2507 A novel multi-photon microscopy method for neuronavigation in deep brain stimulation surgery

2018 ◽  
Vol 2 (S1) ◽  
pp. 2-3
Author(s):  
Nicholas M. George ◽  
Arianna G. Polese ◽  
Greg Futia ◽  
Baris Ozbay ◽  
Wendy Macklin ◽  
...  
Keyword(s):  
Standard Deviation ◽  
Second Harmonic Generation ◽  
Image Classification ◽  
Blood Vessels ◽  
Harmonic Generation ◽  
Calcium Binding ◽  
Second Harmonic ◽  
Brain Regions ◽  
Classification Methods ◽  
Traditional Methods

OBJECTIVES/SPECIFIC AIMS: The goal for this project is to determine the feasibility of using a novel multi-photon fiber-coupled microscope to aid surgeons in localizing STN during surgeries. In order to accomplish this goal, we needed to identify the source of a strong autofluorescent signal in the STN and determine whether we could use image classification methods to automatically distinguish STN from surrounding brain regions. METHODS/STUDY POPULATION: We acquired 3 cadaveric brains from the University of Colorado Anschutz Medical Campus, Department of Pathology. Two of these brains were non-PD controls whereas 1 was diagnosed with PD. We dissected a 10 square centimeter region of midbrain surrounding STN, then prepared this tissue for slicing on a vibratome or cryostat. Samples were immuno-labeled for various cellular markers for identification, or left unlabeled in order to observe the autofluorescence for image classification. RESULTS/ANTICIPATED RESULTS: The border of STN is clearly visible based on the density of a strong autofluorescent signal. The autofluorescent signal is visible using 2-photon (850–1040 nm excitation) and conventional confocal microscopy (488–647 nm excitation). We were also able to visualize blood vessels with second harmonic generation. The autofluorescent signal is quenched by high concentrations of Sudan-black B (0.5%–5%), and is primarily localized in microtubule-associated protein-2 (MAP2)+ cells, indicating that it is likely lipofuscin accumulation in neurons. Smaller lipofuscin particles also accumulate in microglia, identified based on ionized calcium binding adopter 1 (Iba1)+ labeling. We anticipate that colocalization analysis will confirm these qualitative observations. Using 2-photon images of the endogenous autofluorescent signal in these samples, we trained a logistic regression-based image classifier using features derived from gray-level co-occurrence matrices. Preliminary testing indicates that our classifier performed well, with a mean accuracy of 0.89 (standard deviation of 0.11) and a Cohen’s Kappa value of 0.76 (standard deviation of 0.24). We are currently using coherent anti-Stokes Raman scattering and third harmonic imaging to identify different features of myelin that can be used to distinguish between these regions and expect similar results. DISCUSSION/SIGNIFICANCE OF IMPACT: Traditional methods for localizing STN during DBS surgery include the use of stereotactic coordinates and multi-electrode recording (MER) during implantation. MERs are incredibly useful in DBS surgeries, but require penetration of brain structures in order to infer location. Using multi-photon microscopy techniques to aid identification of STN during DBS surgeries offers a number of advantages over traditional methods. For example, blood vessels can be clearly identified with second harmonic generation, something that is not possible with MER. Multi-photon microscopy also allows visualization deep into tissue without actually penetrating it. This ability to look within a depth of field is useful for detection of STN borders based on autofluorescent cell density. When combined with traditional stereotactic information, our preliminary image classification methods are a fast, reliable way to provide surgeons with extra information concerning their location in the midbrain. We anticipate that future advancements and refinements to our image classifier will only increase accuracy and the potential applications and value. In summary, these preliminary data support the feasibility of multi-photon microscopy to aid in the identification of target brain regions during DBS surgeries. The techniques described here complement and enhance current stereotactic and electrophysiological methods for DBS surgeries.

Sign in / Sign up

Export Citation Format

Share Document