Showing papers on "Top-hat transform published in 2018"

MRI and PET image fusion using structure tensor and dual ripplet-II transform

Abstract: Medical image fusion aims at preserving salient image features, reducing the redundancy, and increasing the interpretation quality of images in clinical applications e.g. image-guided surgery. The PET image exhibits functional characteristic with low spatial resolution, while the MRI image exhibits brain tissue anatomy with high spatial resolution. Therefore, the image fusion task is carried out to inject the structural and anatomical information of the high-resolution MRI image into the metabolic information of the PET image. This paper firstly introduces the dual ripplet-II transform (DRT) to overcome the shift variance problem caused by the ripplet-II transform. The proposed transform incorporates the dual-tree complex wavelet into the traditional ripplet-II transform. Secondly, the proposed method takes advantage of the structure tensor and DRT to effectively merge the MRI and PET images. To this end, an objective function is proposed which exploits a weighting matrix to preserve more color and spatial information. Visual and statistical analyses show that the proposed method improves the visual quality and increases the quantitative criteria based on mutual information, edge information, spatial frequency, and structural similarity.

Multi sensor image fusion for surveillance applications using hybrid image fusion algorithm

Abstract: Image fusion is the process of integrating several source images into a single image that provides more reliable information along with reduced redundancy. In this paper, a hybrid image fusion algorithm for multi focus and multi modality images is presented by exploiting the advantages of both the transform as well as spatial domain techniques. In the initial image fusion framework, the source images are decomposed only once using cascaded wavelet transform and the transformed coefficients are combined according to the fusion rules. Inverse cascaded wavelet transform is applied for obtaining the initial fused image. Further, Roberts operator is used for extracting the edge information and decision rule is introduced for choosing the edges from the focused part. The extracted edge information from the focused part replaces the existing edge information in the initial fused image for enhancing the reliability of the fused image. Experiments on various types of images such as multifocus as well as multimodality images are conducted to examine the performance of the proposed algorithm. Experimental results have shown that the proposed algorithm outperforms the well known techniques in terms of both visual perception and quantitative evaluation. Furthermore, the proposed algorithm achieves a good balance between enhancing fusion quality meanwhile reducing the computational cost.

SVM classification of brain images from MRI scans using morphological transformation and GLCM texture features

Abstract: This paper introduces a novel HTT-based GLCM texture feature extraction procedure for an automatic magnetic resonance images (MRI) brain image classification. The method has three phases: 1) hierar...

Segmentation of Retinal Blood Vessel Using Gabor Filter and Extreme Learning Machines

Abstract: The process of obtaining blood vessels from the retinal fundus images plays an important role in the detection of disease in the eye. Analysis of blood vessels provides preliminary information on the presence and treatment of glaucoma, retinopathy, etc. This is why such practices are important. In this study, firstly, features were extracted from color retinal images. Adaptive threshold, Gabor filter and Top-Hat transform were used to make the blood vessel more visible during the feature extraction phase. Subsequently, the acquired features were given as input to the extreme learning machine, and as a result, retinal blood vessel was obtained. At this stage, DRIVE database was used. Twenty colored retinal fundus images were used in the train phase. Thanks to the extreme learning machine, the training process has been carried out quickly (0.42 sec). A high accuracy rate is obtained as %94.59.

Multi-focus image fusion based on nonsubsampled compactly supported shearlet transform

Abstract: Multi-focus image fusion, which aims to combine multi-focus images of a scene to construct an all-in-focus image, has become a major topic in image processing. Different methods have been proposed in spatial or transform domain. But many methods usually suffer from fusion quality degradations, such as contrast reduction, artificial edges, and discontinuous phenomena at boundaries of focused regions, which may cause issues when going for further processing. In order to overcome these problems, we introduce a nonsubsampled compactly supported shearlet transform (NSCSST), which possesses multi-scale, multi-direction, translation invariance and spatial localization characteristics that are very important for image fusion in transform domain. The transform can be implemented sequentially by the shear transform and the separable anisotropic nonsubsampled wavelet transform (SANSWT). Furthermore, we propose a new image fusion method based on NSCSST. It consists of two aspects: multi-direction fusion and transform domain fusion, which respectively correspond to the shear transform and the SANSWT of NSCSST. For each sheared image pair, the SANSWT coefficients are firstly fused by the transform domain fusion rules. And then, the final fused image is obtained by the multi-direction fusion rules, ranging from the simple averaging method to the proposed complex genetic algorithm based method. Experimental results show that our method outperforms some other methods, such as the method based on bilateral gradient, the method based on nonsubsampled contourlet transform, the method based on simultaneous empirical wavelet transform, and the method based on guided filtering.

Digital Image Fusion Using HVS in Block Based Transforms

Abstract: The main aim of image fusion is to integrate the qualitative visual information from multiple images into a single image. Image fusion is implemented in spatial and transform domains. The implementation of algorithm in spatial domain is simple. But, the images are stored/transmitted using popular methods like JPEG and JPEG2000, which are implemented in the transform domain. Therefore fusion algorithms in spatial domain are not suitable for real time application. Image transforms are categorized as block-based and multi resolution-based transforms. In this study, block-based transforms such as Hadamard Transform (HT), Discrete Cosine Transform (DCT), Haar Transform (HrT), and Slant Transform (ST) are considered for image fusion. The DCT based approaches are suffering from undesirable side effects such as blurring and blocking artifacts that reduce the quality of the fused image. In this paper, the Human Visual System (HVS) model is considered to select the appropriate block from multiple images to obtain the fused image. The proposed approach is applied to all the block-based transforms to assess the performance. Methods such as Mutual Information (MI), Edge Strength and Orientation Preservation (ESOP), Feature Similarity Index (FSIM), Normalized Cross Correlation (NCC) and Score are used to assess the performance of the proposed algorithms. The experimental results indicate that the proposed method is better in terms of improved quality and reduced blocking artifacts.

An Adaptive Segmentation Algorithm for Degraded Chinese Rubbing Image Binarization Based on Background Estimation

Abstract: Image Segmentation plays an important role in image processing and analysis. In order to preserve strokes of a Chinese character while enhancing character details for degraded historical document image, we propose an adaptive segmentation algorithm for degraded historical document image binarization based on background estimation for non-uniform illumination images. The novelty of the proposed method is that find an optimal background estimation based on Blind/Referenceless Image Spatial QUality Evaluator. The proposed method has four steps: (i) preprocess using median filtering; (ii) extraction of the red color components; (iii) a morphological operation in order to find an optimal background estimation; and (iv) segmented binary image using Otsu’s Thresholding. Experimental results demonstrate that it is capable of extracting more accurate segmentation of characters for degraded Chinese rubbing document image.

Photovoltaic cell edge breakage and blunt and V-shaped notch detection algorithm

Abstract: The invention provides a photovoltaic cell edge breakage and blunt and V-shaped notch detection algorithm, which is characterized by carrying out image correction and image conversion pretreatment oneach collected frame image to obtain a target image of a photovoltaic cell; carrying out gray top hat transform to obtain an image of the photovoltaic cell only comprising a grid line portion; carrying out threshold segmentation, and filling the grid line portion, so that influence of the internal portion of grid lines on defect detection can be eliminated; obtaining a foreground image only comprising the grid line portion, and carrying out threshold segmentation to obtain a background image; carrying out morphological close operation processing and filling the edge breakage and blunt and V-shaped notch portion; and comparing difference between the background image obtained after morphological processing and the unprocessed background image, so that detection of the edge breakage and bluntand V-shaped notch can be realized. The algorithm overcomes the defect of manual detection, can effectively improve accuracy of defect detection of the photovoltaic cell and has a huge application value for the photovoltaic industry.

Sum of top-hat transform based algorithm for vessel enhancement in MRA images

Abstract: The Magnetic Resonance Angiography (MRA) is rich with information’s. But, they suffer from poor contrast, illumination and noise. Thus, it is required to enhance the images. But, these significant information can be lost if improper techniques are applied. Therefore, in this paper, we propose a new method of enhancement. We applied firstly the CLAHE method to increase the contrast of the image. Then, we applied the sum of Top-Hat Transform to increase the brightness of vessels. It is performed with the structuring element oriented in different angles. The methodology is tested and evaluated on the publicly available database BRAINIX. And, we used the measurement methods MSE (Mean Square Error), PSNR (Peak Signal to Noise Ratio) and SNR (Signal to Noise Ratio) for the evaluation. The results demonstrate that the proposed method could efficiently enhance the image details and is comparable with state of the art algorithms. Hence, the proposed method could be broadly used in various applications.