Robust statistical methods of 2D and 3D image description

Edge characterization has become increasingly important in nanotechnology due to the growing demand for precise nanoscale structure fabrication and assembly. Edge detection and assembly. Edge detection is often performed by thresholding the spatial information of a top-down image obtained by scanning electron microscopy or other surface characterization techniques. Results are highly dependent on an arbitrary threshold value, which makes it difficult to reveal the nature of the real surface and to compare results among images. In this paper, we present an alternative edge boundary detection technique based on the wavelet framework. Our results indicate that the method facilitates nanoscale edge detection and characterization by providing a systematic threshold determination step.

This chapter focuses on the development of novel image enhancement and robust edge detection methods for practical medical image processing. It is known that the popular transformation-domain shrinkage approach for image enhancement applies a fixed mathematical basis to transform all images to be processed for noise or artifact reduction. However it is not adaptable to processed images, and then easily leads to blurring in the enhanced images. On the other hand, the techniques that are commonly used for edge detection are known as gradient and Laplacian operators (or mask), and smoothed gradient masks are typically used for edge detection in noisy images. However, these methods share a common major drawback wherein the associated masks are always fixed irrespective of the noise level in the images. In this study, we propose a novel learning-based method to adaptively deduce the transforming basis or masks from the processing data for medical image enhancement and robust edge detection. By using independent component analysis (ICA), the proposed learning-based method can extract suitable basis functions or masks for image transformation for processing data, which are adaptable to both the processed image and related noise in the image. The efficiency of the proposed learning-based method for medical image enhancement and edge detection is demonstrated experimentally using positron emission tomography (PET) and magnetic resonance imaging (MRI) medical images.

Statistical properties of thresholded images

We propose a novel approach to fully automatic lesion boundary detection in ultrasound breast images. The novelty of the proposed work lies in the complete automation of the manual process of initial Region-of-Interest (ROI) labeling and in the procedure adopted for the subsequent lesion boundary detection. Histogram equalization is initially used to pre-process the images followed by hybrid filtering and multifractal analysis stages. Subsequently, a single valued thresholding segmentation stage and a rule-based approach is used for the identification of the lesion ROI and the point of interest that is used as the seed-point. Next, starting from this point an Isotropic Gaussian function is applied on the inverted, original ultrasound image. The lesion area is then separated from the background by a thresholding segmentation stage and the initial boundary is detected via edge detection. Finally to further improve and refine the initial boundary, we make use of a state-of-the-art active contour method (i.e. gradient vector flow (GVF) snake model). We provide results that include judgments from expert radiologists on 360 ultrasound images proving that the final boundary detected by the proposed method is highly accurate. We compare the proposed method with two existing state-of- the-art methods, namely the radial gradient index filtering (RGI) technique of Drukker et. al. and the local mean technique proposed by Yap et. al., in proving the proposed method's robustness and accuracy.

Significant edges in the case of non-stationary Gaussian noise

We address adaptive detection of extended radar targets embedded in homogeneous Gaussian noise plus subspace interference. We assume that the actual steering vector lies in a fixed and unknown direction of a preassigned and known subspace, while interfering signals are supposed to belong to an unknown subspace, with directions possibly varying from one resolution cell to another. The resulting detection problem is formulated in the framework of statistical hypothesis testing and solved using an ad hoc algorithm strongly related to the GLRT. The performance analysis, carried out also in comparison to natural benchmarks, shows that the proposed detector is an effective solution to the problem of detecting a one-dimensional signal embedded in noise and unknown subspace interference.

We present a parallel architecture for object recognition and location based on concurrent processing of depth and intensity image data. Parallel algorithms for curvature computation and segmentation of depth data into planar or curved surface patches, and edge detection and segmentation of intensity data into extended linear features, are described. Using this feature data in comparison with a CAD model, objects can be located in either depth or intensity images by a parallel pose clustering algorithm. The architecture is based on cooperating stages for low/intermediate level processing and for high level matching. Here, we discuss the use of individual components for depth and intensity data, and their realisation and integration within each parallel stage. We then present an analysis of the performance of each component, and of the system as a whole, demonstrating good parallel execution from raw image data to final pose.

Robust statistical methods are important to the evaluation of toxicological interactions (i.e., departures from additivity) among chemicals in a mixture. However, different concepts of joint toxic action as applied to the statistical analysis of chemical mixture toxicology data or as used in environmental risk assessment often appear to conflict with one another. A unifying approach for application of statistical methodology in chemical mixture toxicology research is based on consideration of change(s) in slope. If the slope of the dose-response curve of one chemical does not change in the presence of other chemicals, then there is no interaction between the first chemical and the others. Conversely, if the rate of change in the response with respect to dose of the first chemical changes in the presence of the other chemicals, then an interaction is said to exist. This concept of zero interaction is equivalent to the usual approach taken in additivity models in the statistical literature. In these additivity models, the rate of change in the response as a function of the i(th) chemical does not change in the presence of other chemicals in a mixture. It is important to note that Berenbaum's (1985, J. Theor. Biol. 114, 413-431) general and fundamental definition of additivity does not require the chemicals in the mixture to have a common toxic mode of action nor to have similarly shaped dose response curves. We show an algebraic equivalence between these statistical additivity models and the definition of additivity given by Berenbaum.

We propose a two stage robust regression procedure to be applied when outliers and leverage points are present. It takes advantage of a high breakdown point MCD estimator and an efficient redescending M-estimator. Its performance was assessed by processing examples from the statistical literature and by carrying out a Monte Carlo simulation experiment; this procedure performs better than other robust regression methods. By pinpointing, labeling and investigating the data generation process of outlying observations, the procedure promotes an interactive attitude in the user and stimulates a thorough scrutiny of data based on subject matter knowledge.

Abstract. This study addresses the question of the existence of a parent flood frequency distribution on a European scale. A new database of L-moment ratios of flood annual maximum series (AMS) from 4105 catchments was compiled by joining 13 national data sets. Simple exploration of the database presents the generalized extreme value (GEV) distribution as a potential pan-European flood frequency distribution, being the three-parameter statistical model that with the closest resemblance to the estimated average of the sample L-moment ratios. Additional Monte Carlo simulations show that the variability in terms of sample skewness and kurtosis present in the data is larger than in a hypothetical scenario where all the samples were drawn from a GEV model. Overall, the generalized extreme value distribution fails to represent the kurtosis dispersion, especially for the longer sample lengths and medium to high skewness values, and therefore may be rejected in a statistical hypothesis testing framework as a single pan-European parent distribution for annual flood maxima. The results presented in this paper suggest that one single statistical model may not be able to fit the entire variety of flood processes present at a European scale, and presents an opportunity to further investigate the catchment and climatic factors controlling European flood regimes and their effects on the underlying flood frequency distributions.

We have developed a new representation of shape in grayscale images that does not rely on the detection of edges. Objects are described by computing features that indicate the geometric stucture of multiscale neighborhoods of each pixel. We define a multiscale medial axis (MMA) by considering the maximum resoponse over scale to the Laplacian operator at each pixel. The MMA is a connected ridgetop stucture. Associated with each pixel on the Mh4A is a width function determined by the scale at which the maximum response to the Laplacian occurs. This width function gives the approximate location of the object’s boundaries. The outputs of multiscale orientation-specific filters give estimates of axis orientation and boundary curvature. This image description has been shown to produce sensible regions for use in image segmentation and image registration applications.

In this article, we reviewed the latest performed works in the image processing using fuzzy systems and methods.First, we introduced the fuzzy processing framework of the images and then study the application of fuzzy methods and systems in the main works of the image processing such as edge detection, image enhancement, segmentation and image description.

The central goal of this dissertation is to design and model a smoothing filter based on the random single and mixed noise distribution that would attenuate the effect of noise while preserving edge details. Only then could robust, integrated and resilient edge detection methods be deployed to overcome the ubiquitous presence of random noise in images. Random noise effects are modeled as those that could emanate from impulse noise, Gaussian noise and speckle noise. In the first step, evaluation of methods is performed based on an exhaustive review on the different types of denoising methods which focus on impulse noise, Gaussian noise and their related denoising filters. These include spatial filters (linear, non-linear and a combination of them), transform domain filters, neural network-based filters, numerical-based filters, fuzzy based filters, morphological filters, statistical filters, and supervised learning-based filters. In the second step, switching adaptive median and fixed weighted mean filter (SAMFWMF) which is a combination of linear and non-linear filters, is introduced in order to detect and remove impulse noise. Then, a robust edge detection method is applied which relies on an integrated process including non-maximum suppression, maximum sequence, thresholding and morphological operations. The results are obtained on MRI and natural images. In the third step, a combination of transform domain-based filter which is a combination of dual tree – complex wavelet transform (DT-CWT) and total variation, is introduced in order to detect and remove Gaussian noise as well as mixed Gaussian and Speckle noise. Then, a robust edge detection is applied in order to track the true edges. The results are obtained on medical ultrasound and natural images. In the fourth step, a smoothing filter, which is a feed-forward convolutional network (CNN) is introduced to assume a deep architecture, and supported through a specific learning algorithm, l2 loss function minimization, a regularization method, and batch normalization all integrated in order to detect and remove impulse noise as well as mixed impulse and Gaussian noise. Then, a robust edge detection is applied in order to track the true edges. The results are obtained on natural images for both specific and non-specific noise-level.

We present a novel multi-flash light field photography (MFLF) technique that couples a light field camera with a ring of four flash lights to combine the benefits of both imaging systems. The multi-flash photography technique enables reliable detection of occlusion edges at individual subaperture images (light field views) in the light field image. We then separately treat occlusion vs. non-occlusion edges in light field stereo matching. For non-occlusion edge pixels, we conduct correspondence matching and depth-from-focus. For pixels on or near the occlusion boundary, we estimate the shape of the occlusion boundary using a novel Support Vector Machine (SVM) approach. We collect an MFLF dataset of synthetic and real scenes that exhibit strong occlusions. The extensive experiments show that our method significantly outperforms the state-of-the-art light field stereo matching methods in accuracy and robustness.

It is very common to use a regular grid like Tian-zi-ge or Mi-zi-ge to help writing in Chinese handwriting environment, especially in education and postal area. Although regular grid is helpful for writing, it is a disaster for recognition. This paper focuses on handwritten Chinese character blind inpainting with regular grid and spot. To solve this problem, we use the recently proposed conditional generative adversarial nets (GANs). Different from the traditional engineering based method like line detection or edge detection, conditional GANs learn a map between target and training data. The generator reconstructs character directly from the data and the discriminator guides the training process to make the generated character more realistic. In this paper, we can automatically remove regular grid in handwritten Chinese character and reconstruct the character's strokes correctly. Moreover, the evaluation on classification task achieved a near state-of-the-art performance on the simulation database and got a convincing result on real world regular grid handwritten Chinese character database.