Multiresolution Image Analysis Research Articles

Relationship between reflectance and degree of polarization in the VNIR-SWIR: A case study on art paintings with polarimetric reflectance imaging spectroscopy

We study the relationship between reflectance and the degree of linear polarization of radiation that bounces off the surface of an unvarnished oil painting. We design a VNIR-SWIR (400 nm to 2500 nm) polarimetric reflectance imaging spectroscopy setup that deploys unpolarized light and allows us to estimate the Stokes vector at the pixel level. We observe a strong negative correlation between the S0 component of the Stokes vector (which can be used to represent the reflectance) and the degree of linear polarization in the visible interval (average -0.81), while the correlation is weaker and varying in the infrared range (average -0.50 in the NIR range between 780 and 1500 nm, and average -0.87 in the SWIR range between 1500 and 2500 nm). By tackling the problem with multi-resolution image analysis, we observe a dependence of the correlation on the local complexity of the surface. Indeed, we observe a general trend that strengthens the negative correlation for the effect of artificial flattening provoked by low image resolutions.

Surface concrete cracks detection and segmentation using transfer learning and multi-resolution image processing

Surface crack detection must be precise and efficient for the structural health monitoring (SHM) of concrete structures. The traditional techniques of crack detection entail human visual examination. These procedures are labor-intensive, expensive, and inefficient. In order to forestall potentially hazardous circumstances caused by concrete surfaces that have deteriorated, it is vital to have an autonomous concrete crack inspection system. Crack detection and segmentation are the two main components of automatic crack inspection. In this research, deep convolutional neural networks (DCNN) and transfer learning (TL) methods are employed for crack detection in concrete structure images, also multiresolution image analysis based on the wavelet transform is the main process in the crack segmentation. The concrete crack dataset is randomly selected from two prevalent concrete crack datasets for developing the proposed method. The results indicated that DCNN classifier models provide good performance with F1-score ranging from 94.5 to 99.6 percent. In addition, the proposed multiresolution image analysis can segment crack pixels with 95.25 percent for the F1-score and shows acceptable and stable performance regarding various segmentation performance metrics.

Multispectral co-occurrence of wavelet coefficients for malignancy assessment of brain tumors.

Brain tumor is not most common, but truculent type of cancer. Therefore, correct prediction of its aggressiveness nature at an early stage would influence the treatment strategy. Although several diagnostic methods based on different modalities exist, a pre-operative method for determining tumor malignancy state still remains as an active research area. In this regard, the paper presents a new method for the assessment of tumor grades using conventional MR sequences namely, T1, T1 with contrast enhancement, T2 and FLAIR. The proposed method for tumor gradation is mainly based on feature extraction using multiresolution image analysis and classification using support vector machine. Since the wavelet features of different tumor subregions, obtained from single MR sequence, do not carry equally important information, a wavelet fusion technique is proposed based on the texture information content of each voxel. The concept of texture gradient, used in the proposed algorithm, fuses the wavelet coefficients of the given MR sequences. The feature vector is then derived from the co-occurrence of fused wavelet coefficients. As each wavelet subband contains distinct detail information, a novel concept of multispectral co-occurrence of wavelet coefficients is introduced to capture the spatial correlation among different subbands. It enables to convey more informative features to characterize the tumor type. The effectiveness of the proposed method is analyzed, with respect to six classification performance indices, on BRATS 2012 and BRATS 2014 data sets. The classification accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and area under curve assessed by the ten-fold cross-validation are 91.3%, 96.8%, 66.7%, 92.4%, 88.4%, and 92.0%, respectively, on real brain MR data.

Optimizing Infrared Camera Resolution for Small Object Detection using Subpixel Rendering and PIFS in Multiresolution Image Analysis.

Background: Breast cancer screening techniques have been developing rapidly in the field of imaging systems. One of these techniques is thermography, which is an alternative modality for mammography to detect breast lesions. Thermography utilization has been progressively developing as various models and methods of object processing improvement. Currently, the Fluke TIS20 infrared camera, with a resolution of 320 × 240, has not been used to measure precisely small objects such as early breast cancer lesions. Retrieval and processing of single images lead into imprecise object measurements and false positive results.Objective: Problems have been arisen due to the limitations of the camera resolution, object retrieval techniques and suboptimal image processing. The aim of this study was to detect accurately breast cancer lesions in rats, which were induced by carcinogenic compounds.Material and Methods: In this experimental study, development of models was conducted based on increasing image by optimizing the ability of low-resolution infrared (IR) cameras to identify s mall objects precisely. Image pixel density increased by adjusting the distance of the objects from the camera and multiple images of objects gradually shifting were used to measure object dimensions precisely.Results: The results showed that cancerous lesions as small as 1.27 mm could be detected. This method of lesion detection had a sensitivity and specificity of 93% and 77 % respectively.Conclusion: Small objects (cancerous lesions) were measured by increasing image resolution through splitting pixels into subpixels and combining several images using Partitioned Iterated Function Systems (PIFS).

Multi-scale convolutional neural network for pixel-wise reconstruction of Van Gogh’s drawings

This paper investigates the reconstruction of Van Gogh’s drawings which have been degraded in the course of time due to aging problems, like ink fading and discoloration. Learning to predict the past and original appearances of degraded drawings can help to envisage how the artist’s work may have looked at the time of creation. In this paper, we use reproductions as reference information for the past appearances of drawings and consider the reconstruction of drawings as a pixel-wise prediction problem. We present an approach to automatically predict the past appearances of drawings. This approach brings together methods from multi-resolution image analysis and deep convolutional neural networks (CNNs) for addressing the task of pixel-wise prediction. Our experiments first investigate how scale affects prediction performance of the proposed multi-scale CNN framework and then demonstrate the reconstruction capability of the multi-scale CNN framework. The results demonstrate that the predictive reconstruction of degraded images is a feasible endeavor.

Digital Image Inpainting by Estimating Wavelet Coefficient Decays From Regularity Property and Besov Spaces

Inpainting methods play a major role in image processing, minimizing the effects of data block loss in image transmission. They analyze primarily the spatial correlation between the portions of an image, usually without explicitly exploring information on the frequency domain. Techniques that evaluate losses of blocks in images do not locally describe a figure of merit for consistent measurement of spatial correlation between pixels of each corrupted block. This investigation presents a digital image inpainting technique, extensible to video applications, using the property of wavelet regularity in multiresolution image analysis, described locally in the sense of Besov vector spaces. Their generalized smoothness allows widespread changes in wavelet coefficients of image data, concealing errors in visual information. This technique estimates decay curves of wavelet coefficients from pixels, based in the observation of the regularity property, correcting pixel blocks contaminated by noise. This investigation proposes an algorithm in the wavelet transform domain, which detects and fixes damaged pixel blocks from a designed decay function for wavelet coefficients. Simulation results, obtained in corrupted images by several noise patterns, indicated better performance of this methodology when compared with other ones.

New FCM Segmentation Approach Based on Multi-Resolution Analysis

This article presents a modified Fuzzy C Means segmentation approach based on multi-resolution image analysis. Fuzzy C-Means standard methods are improved through fuzzy clustering at different image resolution levels by propagating fuzzy membership values pyramidally from a lower to a higher level. Processing at a lower resolution image level provides a rough pixel classification result, thus, a pixel is assigned to a cluster to which the majority of its neighborhood pixels belongs. The aim of fuzzy clustering with multi-resolution images is to avoid pixel misclassification according to the spatial cluster of the neighbourhood of each pixel in order to have more homogeneous regions and eliminate noisy regions present in the image. This method is tested particularly on samples and medical images with gaussian noise by varying multiresolution parameter values for better analysis. The results obtained after multi-resolution clustering are giving satisfactory results by comparing this approach with standard FCM and spatial FCM ones.

Multi-Focus Image Fusion Algorithm in Sensor Networks

Most imaging systems have a limited depth-of-field in the sensor networks that consist of multiple visual sensors. Due to different object distances, not every object can be clearly imaged by a single sensor. This paper proposes a multi-focus image fusion algorithm in sensor networks. The algorithm combines the advantages of multi-scale analysis and image phase analysis. It uses dynamic window and phase stretch transform (PST) to extract the focused regions of each image accurately and performs multi-resolution analysis of images through the non-sub-sampled shearlet transform. Then, the images are fused according to the local standard deviation of PST feature maps. The results of the simulation experiments prove that our algorithm is effective and outperforms some state-of-the-art algorithms.

Three-dimensional multifractal analysis of trabecular bone under clinical computed tomography.

An adequate understanding of bone structural properties is critical for predicting fragility conditions caused by diseases such as osteoporosis, and in gauging the success of fracture prevention treatments. In this work we aim to develop multiresolution image analysis techniques to extrapolate high-resolution images predictive power to images taken in clinical conditions. We performed multifractal analysis (MFA) on a set of 17 exvivo human vertebrae clinical CT scans. The vertebrae failure loads (FFailure) were experimentally measured. We combined bone mineral density (BMD) with different multifractal dimensions, and BMD with multiresolution statistics (e.g., skewness, kurtosis) of MFA curves, to obtain linear models to predict FFailure. Furthermore we obtained short- and long-term precisions from simulated invivo scans, using a clinical CT scanner. Ground-truth data - high-resolution images - were obtained with a High-Resolution Peripheral Quantitative Computed Tomography (HRpQCT) scanner. At the same level of detail, BMD combined with traditional multifractal descriptors (Lipschitz-Hölder exponents), and BMD with monofractal features showed similar prediction powers in predicting FFailure (87%, adj. R2 ). However, at different levels of details, the prediction power of BMD with multifractal features raises to 92% (adj. R2) of FFailure. Our main finding is that a simpler but slightly less accurate model, combining BMD and the skewness of the resulting multifractal curves, predicts 90% (adj. R2) of FFailure. Compared to monofractal and standard bone measures, multifractal analysis captured key insights in the conditions leading to FFailure. Instead of raw multifractal descriptors, the statistics of multifractal curves can be used in several other contexts, facilitating further research.

Segmentation of Breast Lesions in Ultrasound Images through Multiresolution Analysis Using Undecimated Discrete Wavelet Transform.

Earliest detection and diagnosis of breast cancer reduces mortality rate of patients by increasing the treatment options. A novel method for the segmentation of breast ultrasound images is proposed in this work. The proposed method utilizes undecimated discrete wavelet transform to perform multiresolution analysis of the input ultrasound image. As the resolution level increases, although the effect of noise reduces, the details of the image also dilute. The appropriate resolution level, which contains essential details of the tumor, is automatically selected through mean structural similarity. The feature vector for each pixel is constructed by sampling intra-resolution and inter-resolution data of the image. The dimensionality of feature vectors is reduced by using principal components analysis. The reduced set of feature vectors is segmented into two disjoint clusters using spatial regularized fuzzy c-means algorithm. The proposed algorithm is evaluated by using four validation metrics on a breast ultrasound database of 150 images including 90 benign and 60 malignant cases. The algorithm produced significantly better segmentation results (Dice coef = 0.8595, boundary displacement error = 9.796, dvi = 1.744, and global consistency error = 0.1835) than the other three state of the art methods.

SoBT-RFW: Rough-Fuzzy Computing and Wavelet Analysis Based Automatic Brain Tumor Detection Method from MR Images*

One of the important problems in medical diagnosis is the segmentation and detection of brain tumor in MR images. The accurate estimation of brain tumor size is important for treat- ment planning and therapy evaluation. In this regard, this paper presents a new method, termed as SoBT-RFW, for segmentation of brain tumor from MR images. Itintegrates judiciously the merits of rough-fuzzy computing and multiresolution image analysis technique. The proposed method starts with a simple skull stripping algorithm to remove non-cerebral tissues such as skull, scalp, and dura from brain MR images. To extract the scale-space feature vector for each pixel of brain region, the dyadic wavelet analysis is used, while an unsupervised feature selection method, based on maxi- mum relevance-maximum significance criterion, is used to se lect relevant and significant textural features for brain tumor segmentation. To address the uncertainty problem of brain MR image seg- mentation, the proposed SoBT-RFW method uses the robust rough-fuzzyc-means algorithm. After the segmentation process, asymmetricity is analyzed by using the Zernike moments of each of the tissues segmented in the brain to identify the tumor. Finall y, the location of the tumor is searched by a region growing algorithm based on the concept of rough sets. The performance of the proposed SoBT-RFW method, along with a comparison with related approaches, is demonstrated on a set of synthetic and real brain MR images using standard validity indices.

K-means based hybrid wavelet and curvelet transform approach for denoising of remotely sensed images

This article presents a new technique for denoising of remotely sensed images based on multi-resolution analysis (MRA). Multi-resolution techniques provide a coarse-to-fine and scale-invariant decomposition of images for image processing and analysis. The multi-resolution image analysis methods have the ability to analyse the image in an adaptive manner, capturing local as well as global information. Further, noise, as one of the biggest obstacles for image analysis and for further processing, is effectively handled by multi-resolution methods. The article aims at the analysis of noise filtering of image using wavelets and curvelets methods on multispectral images acquired by the QuickBird and medium-resolution Landsat Thematic Mapper satellite systems. To improve the performance of noise filtering, an iterative thresholding scheme and a hybrid approach based on wavelet and curvelet transforms are proposed for restoring the image from its noisy version. Two comparative measures are used for evaluation of the performance of the methods for denoising. One of them is the peak signal-to-noise ratio and the second is the ability of the noise filtering scheme to preserve the sharpness of the edges. By both of these comparative measures, the hybrid approach of curvelet and wavelet for heterogeneous and homogeneous areas with iterative threshold has proved to be better than the others. Results are illustrated using QuickBird and Landsat images for proposed methods and compared with wavelets and curvelet-based denoising.

A Comparative Study Of Different Wavelet Echnic: Denoising The Speckle Noise For Ultrasound Images Using LABVIEW

Biomedical images are generally corrupted by speckle noise and Gaussian noise. Speckle noise is multiplicative type whereas other noises like Gaussian noise are additive type. It is difficult to remove multiplicative noise from images. Latest domain in the field of Image denoising and compression is using wavelet analysis. Multiresolutional image analysis using wavelets is the latest modification in the field of image enhancement and denoising. Speckle Noise is the high frequency content in the ultrasound images and can be easily removed using wavelet based thresholding technique. This paper presents study of various techniques for removal of speckle noise from images, used in biomedical applications, such as Spatial and frequency domain filter and a modified algorithm for speckle noise reduction using wavelet based multiresolutional analysis and thresholding function has been proposed incorporating different wavelets such as Haar, Coiflets, Daubechies and Symlets.

Rough-fuzzy clustering and unsupervised feature selection for wavelet based MR image segmentation.

Image segmentation is an indispensable process in the visualization of human tissues, particularly during clinical analysis of brain magnetic resonance (MR) images. For many human experts, manual segmentation is a difficult and time consuming task, which makes an automated brain MR image segmentation method desirable. In this regard, this paper presents a new segmentation method for brain MR images, integrating judiciously the merits of rough-fuzzy computing and multiresolution image analysis technique. The proposed method assumes that the major brain tissues, namely, gray matter, white matter, and cerebrospinal fluid from the MR images are considered to have different textural properties. The dyadic wavelet analysis is used to extract the scale-space feature vector for each pixel, while the rough-fuzzy clustering is used to address the uncertainty problem of brain MR image segmentation. An unsupervised feature selection method is introduced, based on maximum relevance-maximum significance criterion, to select relevant and significant textural features for segmentation problem, while the mathematical morphology based skull stripping preprocessing step is proposed to remove the non-cerebral tissues like skull. The performance of the proposed method, along with a comparison with related approaches, is demonstrated on a set of synthetic and real brain MR images using standard validity indices.

Rough-fuzzy clustering and multiresolution image analysis for text-graphics segmentation

This paper presents a segmentation method, integrating judiciously the merits of rough-fuzzy computing and multiresolution image analysis technique, for documents having both text and graphics regions. It assumes that the text and non-text or graphics regions of a given document are considered to have different textural properties. The M-band wavelet packet analysis and rough-fuzzy-possibilistic c-means are used for text-graphics segmentation problem. The M-band wavelet packet is used to extract the scale-space features, which offers a huge range of possibilities of scale-space features for document image and is able to zoom it onto narrow band high frequency components. A scale-space feature vector is thus derived, taken at different scales for each pixel in an image. However, the decomposition scheme employing M-band wavelet packet leads to a large number of redundant features. In this regard, an unsupervised feature selection method is introduced to select a set of relevant and non-redundant features for text-graphics segmentation problem. Finally, the rough-fuzzy-possibilistic c-means algorithm is used to address the uncertainty problem of document segmentation. The whole approach is invariant under the font size, line orientation, and script of the text. The performance of the proposed technique, along with a comparison with related approaches, is demonstrated on a set of real life document images.

Noise Filtering of Remotely Sensed Images using Iterative Thresholding of Wavelet and Curvelet Transforms

Abstract. This article presents techniques for noise filtering of remotely sensed images based on Multi-resolution Analysis (MRA). Multiresolution techniques provide a coarse-to-fine and scale-invariant decomposition of images for image interpretation. The multiresolution image analysis methods have the ability to analyze the image in an adaptive manner, capturing local information as well as global information. Further, noise being one of the biggest problems in image analysis and interpretation for further processing, is effectively handled by multi-resolution methods. The paper aims at the analysis of noise filtering of image using wavelets and curvelets on high resolution multispectral images acquired by the Quickbird and medium resolution Landsat Thematic Mapper satellite systems. To improve the performance of noise filtering an iterative thresholding scheme for wavelets and curvelets is proposed for restoring the image from its noisy version. Two comparative measures are used for evaluation of the performance of the methods for denoising. One of them is the signal to noise ratio and the second is the ability of the noise filtering scheme to preserve the sharpness of the edges. By both of these comparative measures, the curvelet with iterative threshold has proved to be better than the others. Results are illustrated using Quickbird and Landsat images for fixed and iterative thresholding.

Compressive sensing image-fusion algorithm in wireless sensor networks based on blended basis functions

Compressive sensing (CS) has given us a new idea at data acquisition and signal processing. It has proposed some novel solutions in many practical applications. Focusing on the pixel-level multi-source image-fusion problem in wireless sensor networks, the paper proposes an algorithm of CS image fusion based on multi-resolution analysis. We present the method to decompose the images by nonsubsampled contourlet transform (NSCT) basis function and wavelet basis function successively and fuse the images in compressive domain. It means that the images can be sparsely represented by more than one basis function. We named this process as blended basis functions representation. Since the NSCT and wavelet basis functions have complementary advantages in multi-resolution image analysis, and the signals are sparser after being decomposed by two kinds of basis functions, the proposed algorithm has perceived advantages in comparison with CS image fusion in wavelet domain which is widely reported by literatures. The simulations show that our method provides promising results.

A New Method Based on Adaptive Discrete Wavelet Entropy Energy and Neural Network Classifier (ADWEENN) for Recognition of Urine Cells from Microscopic Images Independent of Rotation and Scaling

So far, analysis and classification of urine cells number has become an important topic for medical diagnosis of some diseases. Therefore, in this study, we suggest a new technique based on Adaptive Discrete Wavelet Entropy Energy and Neural Network Classifier (ADWEENN) for Recognition of Urine Cells from Microscopic Images Independent of Rotation and Scaling. Some digital image processing methods such as noise reduction, contrast enhancement, segmentation, and morphological process are used for feature extraction stage of this ADWEENN in this study. Nowadays, the image processing and pattern recognition topics have come into prominence. The image processing concludes operation and design of systems that recognize patterns in data sets. In the past years, very difficulty in classification of microscopic images was the deficiency of enough methods to characterize. Lately, it is seen that, multi-resolution image analysis methods such as Gabor filters, discrete wavelet decompositions are superior to other classic methods for analysis of these microscopic images. In this study, the structure of the ADWEENN method composes of four stages. These are preprocessing stage, feature extraction stage, classification stage and testing stage. The Discrete Wavelet Transform (DWT) and adaptive wavelet entropy and energy is used for adaptive feature extraction in feature extraction stage to strengthen the premium features of the Artificial Neural Network (ANN) classifier in this study. Efficiency of the developed ADWEENN method was tested showing that an avarage of 97.58% recognition succes was obtained.

Hierarchical extraction of landslides from multiresolution remotely sensed optical images

The automated detection and mapping of landslides from Very High Resolution (VHR) images present several challenges related to the heterogeneity of landslide sizes, shapes and soil surface characteristics. However, a common geomorphological characteristic of landslides is to be organized with a series of embedded and scaled features. These properties motivated the use of a multiresolution image analysis approach for their detection. In this work, we propose a hybrid segmentation/classification region-based method, devoted to this specific issue. The method, which uses images of the same area at various spatial resolutions (Medium to Very High Resolution), relies on a recently introduced top-down hierarchical framework. In the specific context of landslide analysis, two main novelties are introduced to enrich this framework. The first novelty consists of using non-spectral information, obtained from Digital Terrain Model (DTM), as a priori knowledge for the guidance of the segmentation/classification process. The second novelty consists of using a new domain adaptation strategy, that allows to reduce the expert’s interaction when handling large image datasets. Experiments performed on satellite images acquired over terrains affected by landslides demonstrate the efficiency of the proposed method with different hierarchical levels of detail addressing various operational needs.

Recognition of Tamil Handwritten Characters using Daubechies Wavelet Transforms and Feed-Forward Backpropagation Network

This article suggests wavelet transform based feature extraction technique for extracting robust features from Tamil handwritten characters. The algorithm uses feed-forward back propagation neural network as the classifier. It presents the relevant features of Tamil script and describes various techniques used for character recognition. In common, all the pattern recognition tasks focus on extracting more differentiating features. This is the most important and complicated job. The proposed system concentrates on two dimensional discrete wavelet transformations for extraction of features. For multiresolution analysis of images, Wavelet Transform is used. This capability can be used to study the character image in different frequency bands. Localized basis functions of WT are used for extracting localized features of a character image. This enables us to obtain more distinct traits as features for each character. Feed forward back propagation neural network is one of the general neural network architectures and this architecture can be applied to many different tasks and is very popular.