Iranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11974220171122Retinal Blood Vessel Classification in Fundus Images Based on Structural, Directional and Frequency Features and Optimization with Tagouchi Genetic AlgorithmRetinal Blood Vessel Classification in Fundus Images Based on Structural, Directional and Frequency Features and Optimization with Tagouchi Genetic Algorithm11741107FAGolnoush HamednejadDigital Processing and Machine Vision Research Center, Najafabad Branch, Islamic Azad University and Department of Electrical Eng. of Najafabad Branch, Islamic Azad University.Hossein PourghassemDigital Processing and Machine Vision Research Center, Najafabad Branch, Islamic Azad University and Department of Electrical Eng. of Najafabad Branch, Islamic Azad University.Journal Article20160711Human diseases such as diabetes, high blood pressure and the cerebral source disorders have effects on the retina vessels of human’s eyes. By classifying the retina vessels as two sets of arteries and veins, it can be evaluated the progress and symptoms of mentioned diseases. In this paper, a retinal blood vessel classification algorithm based on structural, directional and frequency features along with feature optimization using Tagouchi genetic algorithm is proposed. For this purpose, to classify the vessels in fundus images, at first the vessels are segmented. In this algorithm, to extract simultaneously information related to direction, diameter and dynamical behavior of the blood vessel, a novel feature based on wavelet transform using entropy contents of DWT and Directional Wavelet Entropy (DWE), Fourier transform using Fourier descriptors have been presented. Also 2-D Frequency Similarity Sectors (2DFSS) is introduced to represent and describe the variations of thickness and direction of the blood vessel. After extracting the feature vector using hybrid model of Genetic algorithm and Tagouchi strategy, the optimal features are selected. Then by employing the multi-layer neural network classifier, the vessels are recognized into arteries and veins classes. With these represented attributes, the classification is performed based on the structure and direction of vessels. Ultimately, the accuracy rate of 82.09% and precision rate of 81.58% are simultaneously obtained in problem of the retinal vessel recognition on a database consisting of 40 images. Human diseases such as diabetes, high blood pressure and the cerebral source disorders have effects on the retina vessels of human’s eyes. By classifying the retina vessels as two sets of arteries and veins, it can be evaluated the progress and symptoms of mentioned diseases. In this paper, a retinal blood vessel classification algorithm based on structural, directional and frequency features along with feature optimization using Tagouchi genetic algorithm is proposed. For this purpose, to classify the vessels in fundus images, at first the vessels are segmented. In this algorithm, to extract simultaneously information related to direction, diameter and dynamical behavior of the blood vessel, a novel feature based on wavelet transform using entropy contents of DWT and Directional Wavelet Entropy (DWE), Fourier transform using Fourier descriptors have been presented. Also 2-D Frequency Similarity Sectors (2DFSS) is introduced to represent and describe the variations of thickness and direction of the blood vessel. After extracting the feature vector using hybrid model of Genetic algorithm and Tagouchi strategy, the optimal features are selected. Then by employing the multi-layer neural network classifier, the vessels are recognized into arteries and veins classes. With these represented attributes, the classification is performed based on the structure and direction of vessels. Ultimately, the accuracy rate of 82.09% and precision rate of 81.58% are simultaneously obtained in problem of the retinal vessel recognition on a database consisting of 40 images. Iranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11974220171122Power Reduction and Fill Factor Improvement on Designing Smart Image Sensor with Motion Detection CapabilityPower Reduction and Fill Factor Improvement on Designing Smart Image Sensor with Motion Detection Capability193441480FAAbbas MahbodDepartment of Electrical and Computer Engineering, University of KashanHossein KarimiyanDepartment of Electrical and Computer Engineering, University of KashanJournal Article20160629Power consumption and fill factor improvement are known as the vitally important parameters in evaluating smart image sensors. In this paper, due to considering two different operation modes in processing different frames, and also designing power management unit, the amount of redundant data processed in unimportant frames has been reduced significantly, and therefore the proposed imaging system consumes less power compared with counterpart imagers. Furthermore, a novel pixel structure is introduced that outputs two consecutive frame voltages in series, with the result that the pixel size is minimized and a higher fill factor is achieved. The performance of this technique is demonstrated using a 64×64 pixel sensor designed in a 0.18µm standard CMOS technology. The sensor chip consumes 0.2mW of power while operating at 100fps with a fill factor of 45%.Power consumption and fill factor improvement are known as the vitally important parameters in evaluating smart image sensors. In this paper, due to considering two different operation modes in processing different frames, and also designing power management unit, the amount of redundant data processed in unimportant frames has been reduced significantly, and therefore the proposed imaging system consumes less power compared with counterpart imagers. Furthermore, a novel pixel structure is introduced that outputs two consecutive frame voltages in series, with the result that the pixel size is minimized and a higher fill factor is achieved. The performance of this technique is demonstrated using a 64×64 pixel sensor designed in a 0.18µm standard CMOS technology. The sensor chip consumes 0.2mW of power while operating at 100fps with a fill factor of 45%.Iranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11974220171122Handwritten word recognition by new feature and lexicon reductionHandwritten word recognition by new feature and lexicon reduction354741729FASomayeh BoroumandDepartment of Computer engineering, Mobarakeh Branch, Islamic Azad UniversityMajid IranpourMobarakehDepartment of Computer engineering and IT, Payam Noor University0000-0003-3009-6093Journal Article20160615Handwritten word recognition (HWR) is very important in document analysis and retrieval. In this paper, an off-line handwritten recognition system for Persian manuscript is introduced. For feature extraction, SIFT descriptors extracted densely from the block of word image and enriched by appending the normalized x and y coordinates and the scale they were extracted at. Linear discriminate analysis (LDA) is used for feature reduction. All words in the dictionary were hierarchically clustered by ISOCLUSE algorithm. In order to recognize the word images, multiple-class and two-class SVM classifiers methods were used. The experimental results showed a better performance in terms of speed and precision of two-class SVM method on the Iranshahr data set. The accuracy of proposed system by select 5 top cluster is shown 93.37% by 76.65% reduction of lexicon.Handwritten word recognition (HWR) is very important in document analysis and retrieval. In this paper, an off-line handwritten recognition system for Persian manuscript is introduced. For feature extraction, SIFT descriptors extracted densely from the block of word image and enriched by appending the normalized x and y coordinates and the scale they were extracted at. Linear discriminate analysis (LDA) is used for feature reduction. All words in the dictionary were hierarchically clustered by ISOCLUSE algorithm. In order to recognize the word images, multiple-class and two-class SVM classifiers methods were used. The experimental results showed a better performance in terms of speed and precision of two-class SVM method on the Iranshahr data set. The accuracy of proposed system by select 5 top cluster is shown 93.37% by 76.65% reduction of lexicon.Iranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11974220171122Two phase MRI segmentation method for brain tumor extractionTwo phase MRI segmentation method for brain tumor extraction495742111FASina TosanMaster of Science Student in Electronic Engineering, Department of Electrical and Computer Eng., University of BirjandNasser MehrshadElectrical and Computer Engineering Department, University of BirjandKazem GhaemiBirjand University of Medical SciencesJournal Article20160214Accurate and timely detection of brain tumor can help patients to choose the kind of treatment and follow-up process and also has a very high impact on the treatment success rate. In this study, a two- phase segmentation method for accurate detection of tumor in the brain magnetic resonance images is provided. In the first stage, after performing the necessary pre-processing schemes on the image, location of the tumor using a threshold-based segmentation method is characterized and secondly, in a marker-based watershed segmentation method is used. Given that in the first stage is not too much emphasis on accurate detection of the tumor area, selecting the threshold values in a wide range of values, will not affect the final results. In the second stage, marker -based segmentation method will lead to accurate detection of the tumor area. The implementation results show that the proposed method in a wide range of input parameters leads to the same accurate results.Accurate and timely detection of brain tumor can help patients to choose the kind of treatment and follow-up process and also has a very high impact on the treatment success rate. In this study, a two- phase segmentation method for accurate detection of tumor in the brain magnetic resonance images is provided. In the first stage, after performing the necessary pre-processing schemes on the image, location of the tumor using a threshold-based segmentation method is characterized and secondly, in a marker-based watershed segmentation method is used. Given that in the first stage is not too much emphasis on accurate detection of the tumor area, selecting the threshold values in a wide range of values, will not affect the final results. In the second stage, marker -based segmentation method will lead to accurate detection of the tumor area. The implementation results show that the proposed method in a wide range of input parameters leads to the same accurate results.Iranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11974220171122Video Based Online Signature VerificationVideo Based Online Signature Verification597343373FABahram AfraM.S. Student of Electrical & Robotic Engineering, Shahrood University OF TechnologyHadi GrailuDept. of Electrical & Robotic Engineering, Shahrood University OF TechnologyJournal Article20160617Signature is one of the personal identification methods. In this paper we have proposed a signature identification approach which is based on using dynamic information of signature video. It composes of four stages. First, the foreground image is extracted. Second the signature shape is obtained by detecting and tracking the pen tip in each frame. The pen tip detection error rate is decreased using a proposed modification. Third, some features based on energy motion image are calculated. At the last stage, the signature model is formed which is used in signature identification procedure. We have generated a signature video database of 50 persons to evaluate the proposed method. The number of 13 signatures of each person are used for training. In addition 8 genuine and 8 fake signatures are used for testing purposes. The Accuracy and Equal Error Rate measures areobtained as 95.02% and 3.8% respectivelySignature is one of the personal identification methods. In this paper we have proposed a signature identification approach which is based on using dynamic information of signature video. It composes of four stages. First, the foreground image is extracted. Second the signature shape is obtained by detecting and tracking the pen tip in each frame. The pen tip detection error rate is decreased using a proposed modification. Third, some features based on energy motion image are calculated. At the last stage, the signature model is formed which is used in signature identification procedure. We have generated a signature video database of 50 persons to evaluate the proposed method. The number of 13 signatures of each person are used for training. In addition 8 genuine and 8 fake signatures are used for testing purposes. The Accuracy and Equal Error Rate measures areobtained as 95.02% and 3.8% respectivelyIranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11974220171122Classification of Benign and Malignant Tumors in Breast Ultrasound Images by using Morphological FeaturesClassification of Benign and Malignant Tumors in Breast Ultrasound Images by using Morphological Features758944400FAHoda NematM.S in Biomedical Engineering, University of Tarbiat ModaresAli MahloojifarDepartment of Biomedical Engineering, Faculty of Electrical and Computer Engineering, University of Tarbiat ModaresAli GooyaDepartment of Biomedical Engineering, Faculty of Electrical and Computer Engineering, University of Tarbiat ModaresNasrin AhmadinejadDepartment of Radiology, Faculty of Medicine, Tehran University of Medical SciencesJournal Article20160801Breast cancer is the second leading cause of death for women all over the world and since the cause of the disease remains unknown, the only method for controlling it is its early detection and diagnosis. The most prominent method for the treatment of breast cancer is biopsy and pathological tests. As the mentioned treatments are invasive and are, in many cases, unnecessary, researchers are in search for high-reliability computer-aided diagnostic systems in order to decrease the number of unnecessary biopsies. These systems consist of four major parts: preprocessing, segmentation, feature extraction and selection, and classification which are beneficial tools for diagnosis of breast cancer. In the present study in order to classify the breast tumors into benign and malignant, borders of the tumors are identified after image preprocessing using with a combination of manual and computerize approaches. In the next stage, 827 features, consisting of 24 shape-based morphological features and 803 border-based morphological features, have been extracted from each image, which 604 of them are recent features added in the present study. Subsequently, a sparse logistic regression classifier was used to eliminate the irrelevant features and classify the images. The data base used in the current study includes 104 Sonography images from breast tumors (72 from benign and 32 from malignant tumors). By applying the suggested algorithm in the present study to images, type of tumors was identified with 89.42% accuracy, 78.13% sensitivity, and 94.44% precision.Breast cancer is the second leading cause of death for women all over the world and since the cause of the disease remains unknown, the only method for controlling it is its early detection and diagnosis. The most prominent method for the treatment of breast cancer is biopsy and pathological tests. As the mentioned treatments are invasive and are, in many cases, unnecessary, researchers are in search for high-reliability computer-aided diagnostic systems in order to decrease the number of unnecessary biopsies. These systems consist of four major parts: preprocessing, segmentation, feature extraction and selection, and classification which are beneficial tools for diagnosis of breast cancer. In the present study in order to classify the breast tumors into benign and malignant, borders of the tumors are identified after image preprocessing using with a combination of manual and computerize approaches. In the next stage, 827 features, consisting of 24 shape-based morphological features and 803 border-based morphological features, have been extracted from each image, which 604 of them are recent features added in the present study. Subsequently, a sparse logistic regression classifier was used to eliminate the irrelevant features and classify the images. The data base used in the current study includes 104 Sonography images from breast tumors (72 from benign and 32 from malignant tumors). By applying the suggested algorithm in the present study to images, type of tumors was identified with 89.42% accuracy, 78.13% sensitivity, and 94.44% precision.