Iranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11978220210723A Comprehensive Survey on DigitalI image Denoising Methods Using Statistical Models in the Transform Domain with the Comparison of ThemA Comprehensive Survey on DigitalI image Denoising Methods Using Statistical Models in the Transform Domain with the Comparison of Them123118217FAMansoore SaeedzarandiIntelligent Data Processing Laboratory (IDPL), Department of Electrical Engineering, Shahid Bahonar, University of Kerman, Kerman, IranHossein Nezamabadi-pourIntelligent Data Processing Laboratory (IDPL), Department of Electrical Engineering, Shahid Bahonar, University of Kerman, Kerman, IranSaeid SaryazdiIntelligent Data Processing Laboratory (IDPL), Department of Electrical Engineering, Shahid Bahonar, University of Kerman, Kerman, IranAhad JamalizadehDepartment of Statistics, Faculty of Mathematics & Computer, Shahid Bahonar University of Kerman,
Kerman, IranJournal Article20201022Image denoising is a well explored topic. Generally, image denoising approaches can be categorized as spatial domain and transform domain methods according to the image representation. Transform domain methods can be divided into two main groups according to their basis functions. Transform domain methods with data adaptive basis functions and transform domain methods with fixed basis functions. Fixed basis functions transform methods, in which, wavelet transform is the most popular, have been widely used for noise reduction applications due to their features and properties, such as frequency / space separation. Also, due to the non-static nature of natural images and the addition of noise to them, statistical methods have received a lot of attention among transform methods. In the present paper, after a brief introduction of denoising methods, the most important statistical models in the fixed basis transform domain are studied. The experimental results are discussed and analyzed to determine the advantages and disadvantages of these methods. The comprehensive study in this paper is a good reference for new research ideas in image denoising.Image denoising is a well explored topic. Generally, image denoising approaches can be categorized as spatial domain and transform domain methods according to the image representation. Transform domain methods can be divided into two main groups according to their basis functions. Transform domain methods with data adaptive basis functions and transform domain methods with fixed basis functions. Fixed basis functions transform methods, in which, wavelet transform is the most popular, have been widely used for noise reduction applications due to their features and properties, such as frequency / space separation. Also, due to the non-static nature of natural images and the addition of noise to them, statistical methods have received a lot of attention among transform methods. In the present paper, after a brief introduction of denoising methods, the most important statistical models in the fixed basis transform domain are studied. The experimental results are discussed and analyzed to determine the advantages and disadvantages of these methods. The comprehensive study in this paper is a good reference for new research ideas in image denoising.https://jmvip.sinaweb.net/article_118217_e5b6deace5d8ff2096cbfbd744230964.pdfIranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11978220210723Stain Normalization of Histopathology Images using conditional Generative Adversarial Networks (cGAN)Stain Normalization of Histopathology Images using conditional Generative Adversarial Networks (cGAN)2541118500FAPegah SalehiImage Processing Research Lab
Dept of Computer Eng. & Info. Tech.
RAZI University, Kermanshah, IRANAbdolah ChalechaleImage Processing Research Lab
Dept of Computer Eng. & Info. Tech.
RAZI University, Kermanshah, IRAN0000-0002-7217-905XJournal Article20201028The diagnosis of cancer is mainly performed by visual analysis of pathologists through examining the morphology of the tissue slices under a microscope. If the microscopic image of a specimen is not stained, it will look colorless and without texture. Therefore, chemical staining is required to create adequate contrast and help identify specific tissue components. During tissue preparation due to differences in chemicals, scanners, and types of illness, similar tissues are usually varied significantly in appearance. This diversity in staining, in addition to interpretive disparity among pathologists, is one of the main challenges in designing robust and flexible systems for automated analysis. Various strategies for stain normalization have been proposed as a pre-processing step in the pipeline of the automated systems. The pix2pix methodwhich is derived from the conditional Generative Adversarial Networks (cGAN) is one of the powerful methods for solving image-to-image translation problems. The main innovation of this paper is to present a new powerful method for the stain normalization of histopathology images using the Pix2Pix method, which is implemented and evaluated on the Mitos-Atypia-14 dataset.In the proposed method, grayscale images are given as input to the network, and then the system learns to restain the texture of the input image in a specific coloring style by preserving the structure and corresponding histopathological pattern. This method, compared to previous methods that relied on a reference image, instead uses the distribution of all images in the learning phase. The proposed method has achieved significant resultsboth in quantitative and qualitative evaluations comparing to some well-known methods in the literature.Moreover, as another innovation, the proposed method tested in a clinical use-case, namely breast cancer tumor classification,using the PatchCamelyon datasetand itshowsa 5% increase in the AUC parameter.The diagnosis of cancer is mainly performed by visual analysis of pathologists through examining the morphology of the tissue slices under a microscope. If the microscopic image of a specimen is not stained, it will look colorless and without texture. Therefore, chemical staining is required to create adequate contrast and help identify specific tissue components. During tissue preparation due to differences in chemicals, scanners, and types of illness, similar tissues are usually varied significantly in appearance. This diversity in staining, in addition to interpretive disparity among pathologists, is one of the main challenges in designing robust and flexible systems for automated analysis. Various strategies for stain normalization have been proposed as a pre-processing step in the pipeline of the automated systems. The pix2pix methodwhich is derived from the conditional Generative Adversarial Networks (cGAN) is one of the powerful methods for solving image-to-image translation problems. The main innovation of this paper is to present a new powerful method for the stain normalization of histopathology images using the Pix2Pix method, which is implemented and evaluated on the Mitos-Atypia-14 dataset.In the proposed method, grayscale images are given as input to the network, and then the system learns to restain the texture of the input image in a specific coloring style by preserving the structure and corresponding histopathological pattern. This method, compared to previous methods that relied on a reference image, instead uses the distribution of all images in the learning phase. The proposed method has achieved significant resultsboth in quantitative and qualitative evaluations comparing to some well-known methods in the literature.Moreover, as another innovation, the proposed method tested in a clinical use-case, namely breast cancer tumor classification,using the PatchCamelyon datasetand itshowsa 5% increase in the AUC parameter.https://jmvip.sinaweb.net/article_118500_eb46ae4e784bf61fdac04116f9a622d7.pdfIranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11978220210723Improving the Accuracy of Natural Dynamic Scenes Recognition using Correlation of Feature Maps in CNNsImproving the Accuracy of Natural Dynamic Scenes Recognition using Correlation of Feature Maps in CNNs4355118502FASafoora HeidariPhd. Student of Electrical Engineering, Ferdowsi University of Mashhad (FUM), Mashhad, IranAbbas Ebrahimi MoghadamDepartment of Electrical Engineering, Faculty of Engineering, Ferdowsi University Of Mashhad (FUM), Mashhad, IranMorteza Khademi DorohDepartment of Electrical Engineering, Faculty of Engineering, Ferdowsi University Of Mashhad (FUM), Mashhad, IranHadi HadizadehDepartment of Electrical Engineering, Quchan University of TechnologyJournal Article20201028Dynamic scene recognition is one of the fundamental research fields in machine vision. In this paper, an effective dynamic scene recognition method using convolutional neural networks is proposed. In the proposed method the correlation of feature maps of different layers in a neural network is exploited as a feature vector containing video information. Firstly, N frames of video are selected and fed into a network to exploit the feature maps, then a Gram matrix indicating the spatial information of the frames of video is calculated. Subsequently, using temporal slicing over selected frames and averaging over the Gram matrices of these frames, temporal information is considered. Encoding features followed by pooling operation, a feature vector is obtained for classification. Experimental evaluations on benchmark dynamic scene datasets demonstrate the effectiveness of the proposed method in comparison with the state-of-the-art methods in this research field and has improved the recognition accuracy about 9% for Maryland dataset and about 3% for YUP++ dataset.Dynamic scene recognition is one of the fundamental research fields in machine vision. In this paper, an effective dynamic scene recognition method using convolutional neural networks is proposed. In the proposed method the correlation of feature maps of different layers in a neural network is exploited as a feature vector containing video information. Firstly, N frames of video are selected and fed into a network to exploit the feature maps, then a Gram matrix indicating the spatial information of the frames of video is calculated. Subsequently, using temporal slicing over selected frames and averaging over the Gram matrices of these frames, temporal information is considered. Encoding features followed by pooling operation, a feature vector is obtained for classification. Experimental evaluations on benchmark dynamic scene datasets demonstrate the effectiveness of the proposed method in comparison with the state-of-the-art methods in this research field and has improved the recognition accuracy about 9% for Maryland dataset and about 3% for YUP++ dataset.https://jmvip.sinaweb.net/article_118502_5ce4c76c8035c5bd2eeba18aa146787b.pdfIranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11978220210723New Grayscale Image Encryption Based on Advanced Encryption Standard and DNA SequenceNew Grayscale Image Encryption Based on Advanced Encryption Standard and DNA Sequence5771118508FAAmirhossein RazmiMsc. Student of Computer Engineering, Islamic Azad University Central Tehran Branch, Tehran, IranKooroush ManochehriComputer Engineering Department,, Amirkabir University of Technology (Garmsar Campus), Tehran, IranAlireza HedayatiDepartment of Computer Engineering, Islamic Azad University of Central Tehran Branch, Tehran, IranJournal Article20201028An image is a visual representation of something that has been created or copied and stored in electronic form. Securing images is becoming an important concern in today’s information security due to the extensive use of images that are either transmitted over a network or stored on disks. Since public media are unreliable and vulnerable to attacks, Image encryption is the most effective way to fulfil confidentiality and protect the privacy of images over an unreliable public media.<br />In this paper a new image encryption algorithm based on Advanced Encryption Standard and DNA sequence is proposed. We present how to encode and decode data in a DNA sequence based on Codon replacement and how to perform the different steps of AES based DNA. The algorithm is implemented in MATLAB 2012b and various performance metrics are used to evaluate its efficacy. The theoretical and experimental analysis show that the proposed algorithm is efficient in speed and precision. Furthermore, the security analysis proves that proposed algorithm has a good resistance against the noise and known attacks; So that Unbreakability of proposed algorithm is 37.48% better than the compared algorithms.An image is a visual representation of something that has been created or copied and stored in electronic form. Securing images is becoming an important concern in today’s information security due to the extensive use of images that are either transmitted over a network or stored on disks. Since public media are unreliable and vulnerable to attacks, Image encryption is the most effective way to fulfil confidentiality and protect the privacy of images over an unreliable public media.<br />In this paper a new image encryption algorithm based on Advanced Encryption Standard and DNA sequence is proposed. We present how to encode and decode data in a DNA sequence based on Codon replacement and how to perform the different steps of AES based DNA. The algorithm is implemented in MATLAB 2012b and various performance metrics are used to evaluate its efficacy. The theoretical and experimental analysis show that the proposed algorithm is efficient in speed and precision. Furthermore, the security analysis proves that proposed algorithm has a good resistance against the noise and known attacks; So that Unbreakability of proposed algorithm is 37.48% better than the compared algorithms.https://jmvip.sinaweb.net/article_118508_31cf6f1f0fd9116d52b63528d7a91b48.pdfIranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11978220210723Classification of retrieved images based on SIFT and Locality Constrained Linear CodingClassification of retrieved images based on SIFT and Locality Constrained Linear Coding7384120264FAMohsen JaberiMsC of Electrical Engineering, Semnan UniversityFarzin YaghmaeeDep. of Electrical and Computer Engineering, Semnan UniversityJournal Article20201209With the growing Internet and digital imaging tools, the size of the image database is increasing rapidly. Therefore, there is a strong need for tools and methods to search for images in a large database. Feature extraction is the most basic step in creating an image-retrieval systems. This paper presents a new method for image retrieval systems. After extracting the feature and computing descriptors for each category by the SIFT algorithm, then the appropriate descriptors are identified by the TF-IDF algorithm and used clustering to find candidate descriptors for each category. In the next step, the descriptor coefficients of each category were used with regard to the representatives from the previous stage by the local coding algorithm as the attribute. Finally we used Extreme Learning Machine (ELM) for classification. Experimental results show that the accuracy achieved in proposed method on the Caltech-101 database is about 98.5% and in Flowers data set is about 97.9%.With the growing Internet and digital imaging tools, the size of the image database is increasing rapidly. Therefore, there is a strong need for tools and methods to search for images in a large database. Feature extraction is the most basic step in creating an image-retrieval systems. This paper presents a new method for image retrieval systems. After extracting the feature and computing descriptors for each category by the SIFT algorithm, then the appropriate descriptors are identified by the TF-IDF algorithm and used clustering to find candidate descriptors for each category. In the next step, the descriptor coefficients of each category were used with regard to the representatives from the previous stage by the local coding algorithm as the attribute. Finally we used Extreme Learning Machine (ELM) for classification. Experimental results show that the accuracy achieved in proposed method on the Caltech-101 database is about 98.5% and in Flowers data set is about 97.9%.https://jmvip.sinaweb.net/article_120264_c5e563f6f9480aeb708998935254621e.pdfIranian Society of Machine Vision and Image ProcessingJournal of Machine Vision and Image Processing2383-11978220210723Blood Vessels Extraction from MRA Images by a Region Growing Algorithm Based on a New Nonlinear Contrast Stretching Function and Shearlets FrameBlood Vessels Extraction from MRA Images by a Region Growing Algorithm Based on a New Nonlinear Contrast Stretching Function and Shearlets Frame8599122739FAMehdi MirzafamPhD Student, Azarbaijan Shahid Madani University, TabrizNasser AghazadehImage Processing Laboratory, Department of Applied Mathematics, Azarbaijan Shahid Madani University, Tabriz, IranJournal Article20210118Region growing, in a simple version, is a segmentation process, which having pixels as seeds, pixels with the same intensities and connected them are added to the area gradually, and finally presents a binary image that contains the object or objects of the target. So far, many binary segmentation techniques have been developed to extract target objects, with the common disadvantage that they do not perform the extraction task completely. Frames as the generalization of orthogonal bases are used scarcely in these algorithms. In this paper, a new nonlinear contrast stretching function is introduced, and then, based on the contrast stretching function andshearlets frame, correct initialization seeds are extractedand then the region growing algorithm applyto the image. The results presented on synthetic images and real medical images show the advantages of our technique to those recently proposed.Region growing, in a simple version, is a segmentation process, which having pixels as seeds, pixels with the same intensities and connected them are added to the area gradually, and finally presents a binary image that contains the object or objects of the target. So far, many binary segmentation techniques have been developed to extract target objects, with the common disadvantage that they do not perform the extraction task completely. Frames as the generalization of orthogonal bases are used scarcely in these algorithms. In this paper, a new nonlinear contrast stretching function is introduced, and then, based on the contrast stretching function andshearlets frame, correct initialization seeds are extractedand then the region growing algorithm applyto the image. The results presented on synthetic images and real medical images show the advantages of our technique to those recently proposed.https://jmvip.sinaweb.net/article_122739_f1506bf0647585566b77de25abfd1c5f.pdf