Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Extracting the combination featuresbased on the binary genetic to improve the diagnostic of iris recognition systems
1
11
FA
Mehran
Nasrpour
Mcs. Student, Computer Engineering Department, Hamedan Branch, Islamic Azad University
nasrpour@gmail.com
Mansour
Esmaeilpour
Computer Engineering Department, Hamedan Branch, Islamic Azad University
esmaeilpour@iauh.ac.ir
Iris recognition system consists of some stages where feature extraction is one of the most important one. Most of the available systems uses one special technic to extract features. To improve performance of the system, we used a binary genetic algorithm with a novel fitness function to find a combinational feature extraction method. Proposed method uses many different filters and transformations which were applied in feature extraction of iris and finds the best combination during iteration of the algorithm. Consequently a set of methods including numbers of wavelet transform, Gabor filter and Fourier transform is achieved as the best combinational feature extraction approach. In experiments, improving performance of the proposed combinational approach is shown contrasting to other single methods using ROC curve. Comparisons showed that the proposed method outperforms other state of the art methods in most of cases. This method succeeded to achieve FAR=0 and FRR=0.092.
Biometrics,IRIS,Identification,Feature Extraction and Genetic Algorithm
https://jmvip.sinaweb.net/article_50582.html
https://jmvip.sinaweb.net/article_50582_2af54df1b53e1aa93fb15df405d958b7.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Tumor segmentation in mammogram images using Chan-Vese active contour and texture local feature information
13
25
FA
Fatemeh
Shirazi
M.S. Student Department of Electrical Engineering, Graduate University of Advanced Technology
shirazi.fatemeh0@gmail.com
Esmat
Rashedi
Department of Electrical Engineering, Graduate University of Advanced Technology
e.rashedi@kgut.ac.ir
Hossein
Nezamabadi-pour
Department of electrical engineering, Shahid Bahonar University of Kerman
nezam@uk.ac.ir
Cancerous tumor segmentation in mammogram images is an important stage and a challenging problemin computer aided detection (CAD) systems. In this paper, local feature information and Chan-Vese(LFI-CV)active contour modelare used for tumor segmentation. First, the texture feature mapsof mammograms are extracted. The utilized texture feature information includes gray level co-occurrence matrix (GLCM) and Gabor features. Using this information,the force values ofChan-Vese model are set and active contour model’s energy is minimized.As a result, the contour accurately segments the tumor. The results show that tumor segmentation using the proposed active contour modelandGabor texture feature at orientationis efficient in regard to the number of iterations, accuracy, and sensitivity. The mini-MIAS database is used for evaluation.
breast cancer,Computer aided detection,mammography,Cancerous tumor segmentation,Texture feature,Local feature information,Chan-Vese active contour model
https://jmvip.sinaweb.net/article_51045.html
https://jmvip.sinaweb.net/article_51045_64c153aee55cda516076a66130e565e0.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Brain tumors detection by combination of adaptive neuro-fuzzy inference system and hierarchical clustering
27
37
FA
Abdolhossein
Fathi
Department of Computer Engineering, Faculty of Engineering, Razi University, Kermanshah
a.fathi@razi.ac.ir
mehdi
taheri
MSc Student, Department of Computer Engineering, Faculty of Engineering, Razi University, Kermanshah
mehdi.taheri2049@gmail.com
Fardin
Abdali-Mohammadi
Department of Computer Engineering, Faculty of Engineering, Razi University, Kermanshah
fardin.abdali@razi.ac.ir
Detection of brain tumors region is a crucial step in automatic detection and treatment systems. This paper presents a hybrid method based on adaptive neuro-fuzzy inference system (ANFIS) and hierarchical clustering to identify location and region of brain tumors. For this purpose, first the center line of brain is detected, and then brain region is divided into non-overlapped blocks. Then, for each block intensity and texture features are extracted. With exploitation symmetry features of two hemispheres of the brain, blocks containing tumor tissue are recognized using ANFIS classifier. Finally by smoothing brain MRI image and exploiting hierarchical clustering, exact region of tumor is specified. The proposed method was tested on Harvard MRI dataset. The obtained performance of the proposed method with criterions accuracy, sensitivity and specificity are 98.1±4.7%, 94.1±3.2% and 98.7±4.9% respectively.
Tumor detection,Neuro-fuzzy inference,Hierarchical clustering,Segmentation,Magnetic resonance images
https://jmvip.sinaweb.net/article_51469.html
https://jmvip.sinaweb.net/article_51469_c211c6bb2c07ea86a64bf84066d8f60b.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
A Review on Image Registration Methods, Concepts and applications
39
67
FA
Zahra
Hossein-Nejad
Department of Electrical Engineering, Sirjan Branch, Islamic Azad University, Sirjan, Iran
hoseinnejad.zahra@yahoo.com
Mehdi
Nasri
0000-0002-9254-3584
Department of Electrical Engineering, Faculty of Engineering, Azad University Khomeinishahr Branch (Isfahan)
nasri_me@iaukhsh.ac.ir
Image registration is one of the fields widely used in image processing where much research has been done. Image registration is thealignment and compliance of two or more images from different imaging conditions. Itsapplications include change identification between images, image fusion, object recognition, and image mosaic. In this paper, in addition to introducing the concepts of image registration, we have collected and classified different researches, as well as definition of the research approach thereof. Moreover, thevarious aspects of image registration have been evaluated through four different tests. This paper could pave the way for researchers in the field of image processing as and it has been tried to includeall aspects of this field of study herein.
Image Registration,feature detection,Matching,transform modal estimation,SIFT algorithm
https://jmvip.sinaweb.net/article_51831.html
https://jmvip.sinaweb.net/article_51831_ca6db7ca89a92f4926cc71a546f8cc1a.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
An intelligent hybrid method for the diagnosis, segmentation and classification of breast tumors based on new tissue features extracted from two views of mammography images
69
83
FA
Nooshin
Bigdeli
Department of Electrical Engineering, Imam Khomeini International University Qazvin
n.bigdeli@eng.ikiu.ac.ir
Hamed
Jabbari
PhD Student, of Electrical Engineering, Imam Khomeini International University Qazvin
h_jabbari@edu.ikiu.ac.ir
Negar
Maleki
Department of Electrical Engineering, Imam Khomeini International University Qazvin
nmaleki77@yahoo.com
Breast cancer is one of the most important cancers among women. Usually, screening for breast cancer is mammography, which reduced the death rate caused by it. The purpose of this paper is to introduce a new hybrid intelligent method for classification of breast tissue into two healthy and unhealthy types by simultaneous examination of two aspects of mammogram images and segmentation of unplanned unhealthy tissue. To this purpose, a new hybrid method including clustering and region growth algorithms are used to identification of the suspected area to the tumor presence. The suspected area is identifiedby combining the FCM clustering and the region growth algorithms after removed the background, and was segmented tumor using the morphological processes. Then, was done classification of the breast tissue into two types of healthy and unhealthyusingsimultaneous two standard views of mammogram (MLO and CC) of a breast, and the extraction of tissue features based on the gray-level co-occurrence matrix,and c and the brightest of intensity level of the cluster center features.Also, was introduced and used the brightest of intensity level of the cluster center features for the first time. Finally, the extracted features are considered as inputs of a fuzzy system for classification of breast tissue. The results of this study are shown the proposed method has accuracy 97.7% inthe breast tissue classification on 300 pairs of mammograms. Also, it is shown that the simultaneous examination of features of the two views standard mammograms can be useful in early detection of breast cancer.
breast cancer,Breast tissue classification,Growing area algorithm,Feature Extraction,Fuzzy inference system
https://jmvip.sinaweb.net/article_52052.html
https://jmvip.sinaweb.net/article_52052_852b7c223fad9b9b248ac766ece54d5e.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Prediction and Control of the Water Content of the Turfgrass Plant by an Intelligent System Using Image Processing and Support Vector Regression Algorithm
85
102
FA
Maryam
Nadafzadeh
MSc. Student, Department of Mechanics of Biosystems Engineering, Faculty of Agricultural and Rural Development, Ramin Agriculture and Natural Resources University of Khuzestan, Ahvaz, Khuzestan Iran
maryam.nadaf@yahoo.com
Saman
Abdanan Mehdizadeh
Mechanics of Biosystems Engineering Department, Faculty of Agricultural Engineering and Rural Development, Ramin Agriculture and Natural Resources University of Khuzestan
saman.abdanan@gmail.com
Mohammadreza
Salehi Salmi
Horticultural Science Department, Faculty of Agriculture, Ramin Agriculture and Natural Resources University of Khuzestan
m_salehisalmi@yahoo.com
Due to the environmental changes and increasing global temperature and drought conditions, irrigation of plants along with theirs protection of growth as well as high yield is very important. Therefore, water consumption can be reduced significantly in agriculture by monitoring and control of plant growth conditions and increase the irrigation efficiency through the conversion of surface irrigation methods to smart irrigation systems in the situation of water crisis. In this study, to detect plant water requirement, a set of turf grass plants images were taken and studied under drought stress conditions to extract the color, texture and number of features in the frequency domain. Thereafter, all the extracted parameters of images were investigated, then, according to the results of statistical analysis (p<0.05), most appropriate features were selected to predict water content of plant by support vector regression algorithm (SVR). Finally, it was shown that the linear kernel function of SVR algorithm has the highest correlation coefficient (0.95) and the lowest values of MAPE (14.08), RMSE (0.10), SRE (0.063) and RAV (0.14) compared to other kernels. Thus, this indicated that the ability of suggested system to measure and evaluate wilting plant conditions and control of required water for plant.
Digital images processing,Turfgrass,Intelligent irrigation control,Support vector regression
https://jmvip.sinaweb.net/article_52256.html
https://jmvip.sinaweb.net/article_52256_3d704d7cffb646f346ae16e2e33efaaf.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Estimation of The Location and Orientation of a Camera in Short Image Sequence Using Extended Kalman Filter
103
113
FA
Mohammad Amin
Mehralian
PhD Candidate, Artificial Intelligence and Robotic, Iran University of Science and Technology
mehralian@comp.iust.ac.ir
Mohsen
Soryani
0000-0002-8555-9617
Department of Computer Engineering, Iran University of Science and Technology
soryani@iust.ac.ir
Estimating the camera location and orientation from an image sequence is wildly used in many applications such as Augmented Reality and Robot Navigation. In this paper a new hybrid method is proposed. For this purpose, we apply Extended Kalman Filter to estimate the camera trajectory with 6 degree of freedom. The main difference between the proposed method and other filter-based methods is that in the proposed method, 3D points of the world model have been removed from the state vector and alternatively, Multiple View Geometry methods have been used to initialize the world model without uncertainty. As a result, the complexity of the algorithm, which is one of the shortcomings of the filter-based methods, has been reduced. Computer simulations and experimental results show that the proposed method improves pose estimation compared to PnP methods.
3D Computer Vision,Camera Pose Estimation,Multiple View Geometry,Extended Kalman Filter
https://jmvip.sinaweb.net/article_54562.html
https://jmvip.sinaweb.net/article_54562_faa169f5c252c2481b521315744a5a16.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Tasvirnet Image Network(Experiences,challenges and localization tools for ImageNet)
115
126
FA
Farzad
Zargari
0000-0003-1585-9283
Faculty member, Iran Telecom Research Center, Tehran, Iran
zargari@itrc.ac.ir
farzaneh
Rahmani
IT platforms department, IT faculty, Iran Telecommunication Research Center, Tehran, Iran
rahmani@itrc.ac.ir
Mojhgan
Farhoodi
IT platforms department, IT faculty, Iran Telecommunication Research Center, Tehran, Iran
farhoodi@itrc.ac.ir
Mohammad Hossein
Zabolzadeh
IT faculty, Iran Telecommunication Research Center, Tehran, Iran
zabolzadeh@itrc.ac.ir
Zeinab
Porkar
IT platforms department, IT faculty, Iran Telecommunication Research Center, Tehran, Iran
z.porkar@itrc.ac.ir
Ehsan
Ghasemi
IT platforms department, IT faculty, Iran Telecommunication Research Center, Tehran, Iran
e.ghasemi@itrc.ac.ir
Large image databases are used as training datain visual artificial intelligence applications and deep learning algorithms. Tasvirnet image network is a hierarchical image database in accordance with Iranian and Islamic culture that provides about 8 million images for over 30,000 words.Synset hierarchy of Tasvirnet is based on hierarchy of ImageNet anditssynsets are translated to Persian using automatic translation. There are 7890745 images in Tasvirnet (for 32295 Persian Synsets) which are collected using auromatic downloading of image links provided by ImageNet. Moreover,71873 images for up to 1000 synsets related to Iranian and Islamic Culture are prepared using crowdsourcing method and they are added to Tasvirnet. The purpose of this paper is transferring of experiences, production challenges, and developed tools to create a hierarchical image database called Tasvirnet, as well as an outline of the crowdsourcing process used to provide the extraimages for Iranian and Islamic Culture. based synsets.
Hierarchical Image Databases,Persian Image Network,Crowdsourcing
https://jmvip.sinaweb.net/article_57367.html
https://jmvip.sinaweb.net/article_57367_6820878c34fa592e67dc1b6f2e71a467.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
Mitosis Detection in breast cancer biopsy images using Extreme Learning Machines
127
138
FA
Sooshiant
Zakariapor
Ms.C Student, Department of Electrical and Computer Engineering, Babol Noshirvani University of Technology.
sooshiant@stu.nit.ac.ir
Hamid
Jazayeriy
Department of Computer Engineering, Faculty of Electrical and Computer Engineering, Babol Noshirvani University of Technology.
jhamid@nit.ac.ir
Mehdi
Ezoji
Department of Electrical and Computer Engineering, Babol Noshirvani University of Technology.
m.ezoji@nit.ac.ir
Counting mitotic cellsis one of the main tasks involved in assessing breast cancer proliferation grade. Unfortunately, detection of mitoses present in the tissue is a challenging task. These cells have a wide variety of shape configurations and are sometimes very similar to apoptotic cells or external objects in the tissue sample. Utilizing image processing for automatic detection of mitotic cells is likely to reduce human errorandincreasegrading speed and performance.Most available mitosis detection methods extract many features from cells then classify cells using classic classifiers, or else, directly classify cells using neural networks. The former are fast but inaccurate methods, the latter being slow but accurate. In this work, we aim to present a simultaneously fast and accurate method based on a special type of neural networks, called ELM. After a pre-processing step, candidate cells are selected using thresholding and finding local maxima. An ELM is then directly trained with each cell image, without feature extraction. Our results indicate a considerable improvement over the status-quo. Our method also benefits from a very fast training time and test time.
Extreme Learning Machines,breast cancer,Mitosis detection,Digital Pathology,Biopsy Images,Cell classification
https://jmvip.sinaweb.net/article_57680.html
https://jmvip.sinaweb.net/article_57680_d9336b87fa7a494fd31a9e3f6f61b48d.pdf
Iranian Society of Machine Vision and Image Processing
Journal of Machine Vision and Image Processing
2383-1197
5
2
2018
11
22
A review on computer aided systems for mass detection using 3D automatic breast ultrasound (ABUS) images
139
153
FA
Ehsan
Kozegar
Phd Student, Iran University of Science and Technology
e_koozegar@comp.iust.ac.ir
Mohsen
Soryani
0000-0002-8555-9617
Dept. of Computer Eng. Iran University of Science and Technology
soryani@iust.ac.ir
Hamid
Behnam
Dept. of Electrical Eng. Iran University of Science and Technology
behnam@iust.ac.ir
Masoumeh
Salamati
Department of Reproductive Imaging, Reproductive Biomedicine Research Center, Royan Institute for Reproductive Biomedicine, ACECR
dr.m.salamati@gmail.com
Three dimensional Automatic Breast UltraSound (ABUS) is a modern and effective imaging system which can be used as an adjunct to mammography for women with dense breasts. In this paper, the ABUS imaging system is introduced and its advantages over current handheld ultrasound and other modalities are compared. Then, we emphasize on the benefits of the computerized systems to detect masses in 3D ultrasound images of whole breasts. Consequently, different state-of-the-art computer aided mass detection systems for this type of images are described in details and their limitations are discussed. Finally, potential solutions to overcome these limitations are presented.
Automatic Breast UltraSound Imaging,Mass Detection,Computer Aided Systems,image processing
https://jmvip.sinaweb.net/article_58060.html
https://jmvip.sinaweb.net/article_58060_618b09af1f7717033e95ab1cea4b2035.pdf