Dana Elena Ilea

Image segmentation based on the integration of colour-texture descriptors-A review

The adaptive integration of the colour and texture attributes in the development of complex image descriptors is one of the most investigated topics of research in computer vision. The substantial interest shown by the research community in colour-texture-based segmentation is mainly motivated by two factors. The first is related to the observation that the imaged objects are often described at perceptual level by distinctive colour and texture characteristics, while the second is motivated by the large spectrum of possible applications that can be addressed by the colour-texture integration in the segmentation process. Over the past three decades a substantial number of techniques in the field of colour-texture segmentation have been reported and it is the aim of this article to thoroughly evaluate and categorise the most relevant algorithms with respect to the modality behind the integration of these two fundamental image attributes. In this paper we also provide a detailed discussion about data collections, evaluation metrics and we review the performance attained by state of the art implementations. We conclude with a discussion that samples our views on the field of colour-texture image segmentation and this is complemented with an examination of the potential future directions of research.

CTex—An Adaptive Unsupervised Segmentation Algorithm Based on Color-Texture Coherence

This paper presents the development of an unsupervised image segmentation framework (referred to as CTex) that is based on the adaptive inclusion of color and texture in the process of data partition. An important contribution of this work consists of a new formulation for the extraction of color features that evaluates the input image in a multispace color representation. To achieve this, we have used the opponent characteristics of the RGB and YIQ color spaces where the key component was the inclusion of the self organizing map (SOM) network in the computation of the dominant colors and estimation of the optimal number of clusters in the image. The texture features are computed using a multichannel texture decomposition scheme based on Gabor filtering. The major contribution of this work resides in the adaptive integration of the color and texture features in a compound mathematical descriptor with the aim of identifying the homogenous regions in the image. This integration is performed by a novel adaptive clustering algorithm that enforces the spatial continuity during the data assignment process. A comprehensive qualitative and quantitative performance evaluation has been carried out and the experimental results indicate that the proposed technique is accurate in capturing the color and texture characteristics when applied to complex natural images.

Fully automated segmentation and tracking of the intima media thickness in ultrasound video sequences of the common carotid artery

The robust identification and measurement of the intima media thickness (IMT) has a high clinical relevance because it represents one of the most precise predictors used in the assessment of potential future cardiovascular events. To facilitate the analysis of arterial wall thickening in serial clinical investigations, in this paper we have developed a novel fully automatic algorithm for the segmentation, measurement, and tracking of the intima media complex (IMC) in B-mode ultrasound video sequences. The proposed algorithm entails a two-stage image analysis process that initially addresses the segmentation of the IMC in the first frame of the ultrasound video sequence using a model-based approach; in the second step, a novel customized tracking procedure is applied to robustly detect the IMC in the subsequent frames. For the video tracking procedure, we introduce a spatially coherent algorithm called adaptive normalized correlation that prevents the tracking process from converging to wrong arterial interfaces. This represents the main contribution of this paper and was developed to deal with inconsistencies in the appearance of the IMC over the cardiac cycle. The quantitative evaluation has been carried out on 40 ultrasound video sequences of the common carotid artery (CCA) by comparing the results returned by the developed algorithm with respect to ground truth data that has been manually annotated by clinical experts. The measured IMTmean ± standard deviation recorded by the proposed algorithm is 0.60 mm ± 0.10, with a mean coefficient of variation (CV) of 2.05%, whereas the corresponding result obtained for the manually annotated ground truth data is 0.60 mm ± 0.11 with a mean CV equal to 5.60%. The numerical results reported in this paper indicate that the proposed algorithm is able to correctly segment and track the IMC in ultrasound CCA video sequences, and we were encouraged by the stability of our technique when applied to data captured under different imaging conditions. Future clinical studies will focus on the evaluation of patients that are affected by advanced cardiovascular conditions such as focal thickening and arterial plaques.

An automatic 2D CAD algorithm for the segmentation of the IMT in ultrasound carotid artery images

Common carotid intima-media thickness (IMT) is a reliable measure of early atherosclerosis — its accurate measurement can be used in the process of evaluating the presence and tracking the progression of disease. The aim of this study is to introduce a novel unsupervised Computer Aided Detection (CAD) algorithm that is able to identify and measure the IMT in 2D ultrasound carotid images. The developed technique relies on a suite of image processing algorithms that embeds a statistical model to identify the two interfaces that form the IMT without any user intervention. The proposed image segmentation scheme is based on a spatially continuous vascular model and consists of several steps including data preprocessing, edge filtering, model selection, edge reconstruction and data refinement. To conduct a quantitative evaluation each image was manually segmented by clinical experts and performance metrics between the segmentation results obtained by the proposed method and the ground truth data were calculated. The experimental results show that the proposed CAD system is robust in accurately estimating the IMT in ultrasound carotid data.

Local binary patterns versus signal processing texture analysis: a study from a performance evaluation perspective

Purpose – The purpose of this paper is to review and provide a detailed performance evaluation of a number of texture descriptors that analyse texture at micro‐level such as local binary patterns (LBP) and a number of standard filtering techniques that sample the texture information using either a bank of isotropic filters or Gabor filters.Design/methodology/approach – The experimental tests were conducted on standard databases where the classification results are obtained for single and multiple texture orientations. The authors also analysed the performance of standard filtering texture analysis techniques (such as those based of LM and MR8 filter banks) when applied to the classification of texture images contained in standard Outex and Brodatz databases.Findings – The most important finding resulting from this study is that although the LBP/C and the multi‐channel Gabor filtering techniques approach texture analysis from a different theoretical perspective, in this paper the authors have experimentall...

Texture Enhanced Histogram Equalization Using TV-

Histogram transformation defines a class of image processing operations that are widely applied in the implementation of data normalization algorithms. In this paper, we present a new variational approach for image enhancement that is constructed to alleviate the intensity saturation effects that are introduced by standard contrast enhancement (CE) methods based on histogram equalization. In this paper, we initially apply total variation (TV) minimization with a L1 fidelity term to decompose the input image with respect to cartoon and texture components. Contrary to previous papers that rely solely on the information encompassed in the distribution of the intensity information, in this paper, the texture information is also employed to emphasize the contribution of the local textural features in the CE process. This is achieved by implementing a nonlinear histogram warping CE strategy that is able to maximize the information content in the transformed image. Our experimental study addresses the CE of a wide variety of image data and comparative evaluations are provided to illustrate that our method produces better results than conventional CE strategies.

{\rm L}^{1}

One important step in the process of colour image segmentation is to reduce the errors caused by image noise and local colour inhomogeneities. This can be achieved by filtering the data with a smoothing operator that eliminates the noise and the weak textures. In this regard, the aim of this paper is to evaluate the performance of two image smoothing techniques designed for colour images, namely bilateral filtering for edge preserving smoothing and coupled forward and backward anisotropic diffusion scheme (FAB). Both techniques are non-linear and have the purpose of eliminating the image noise, reduce weak textures and artefacts and improve the coherence of colour information. A quantitative comparison between them will be evaluated and also the ability of such techniques to preserve the edge information will be investigated.

Image Decomposition

Statistical partitioning of images into meaningful areas is the goal of all region-based segmentation algorithms. The clustering or creation of these meaningful partitions can be achieved in number of ways but in most cases it is achieved through the minimization or maximization of some function of the image intensity properties. Commonly these optimization schemes are locally convergent, therefore initialization of the parameters of the function plays a very important role in the final solution. In this paper we perform an automatically initialized expectation-maximization algorithm to partition the data in medical MRI images. We present analysis and illustrate results against manual initialization and apply the algorithm to some common medical image processing tasks.

Adaptive Pre-Filtering Techniques for Colour Image Analysis

Data smoothing and feature enhancement are two important precursors to many higher-level computer vision applications such as image segmentation and scene understanding. Total variation (TV) flow algorithms are a distinct subcategory of diffusion-based filtering techniques that have been widely applied to reduce the level of noise in the image but not at the expense of poor feature preservation. In this paper we address a number of numerical aspects associated with the TV flow and in particular we are interested to redefine the TV flow regularization in order to reduce the effect of oscillations and improve the convergence of the implementations in the discrete domain. TV flow algorithms are implemented using iterative schemes and one difficult problem is the selection of appropriate criteria to identify the optimal number of iterations. In this paper we show that the application of a time-ageing procedure leads to an elegant formulation were the TV flow algorithms converge naturally to the optimal solution. To evaluate the performance of the proposed algorithm (referred in this paper to as time-controlled (TC)-TV flow), a large number of experiments on various types of natural images were conducted.

Automatic seed initialization for the expectation-maximization algorithm and its application in 3D medical imaging

In this paper we present a parameter optimisation procedure that is designed to automatically initialise the number of clusters and the initial colour prototypes required by data space partitioning techniques. The proposed optimisation approach involves a colour saliency measure used in conjunction with a SOM classification procedure. For evaluation purposes, we have integrated the proposed initialisation technique in an unsupervised colour segmentation scheme based on K-Means clustering and the evaluation has been carried out in the context of the unsupervised segmentation of natural images.