An l ½ and Graph Regularized Subspace Clustering Method for Robust Image Segmentation

Due to the presence of diverse and disparate patterns, the segmentation of natural images endures crucial as well as a challenging problem in image analysis algorithms. The proposed method addresses image segmentation as a subspace clustering of image feature vectors. Initially, an image is partitioned into superpixels and further, a feature data matrix is computed using the Local Spectral Histogram (LSH) features from individual superpixels. A single-stage optimization model is formulated which incorporates better subspace selection, excellent grouping effect and simultaneous noise robustness for the uncorrelated, correlated and corrupted data by the conjunctive venture of l <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> , l <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> and l <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2,1</sub> norm minimization. The proposed model is solved using Augmented Lagrangian technique. We compared the proposed method with state-of-the-art methods and the results demonstrate the improved performance of our proposed model over the existing counterparts.

Grouping points by shared subspaces for effective subspace clustering

Image segmentation by correlation adaptive weighted regression

In this paper, we present a new image segmentation method based on the concept of sparse subset selection. Starting with an over-segmentation, we adopt local spectral histogram features to encode the visual information of the small segments into high-dimensional vectors, called superpixel features. Then, the superpixel features are fed into a novel convex model which efficiently leverages the features to group the superpixels into a proper number of coherent regions. Our model automatically determines the optimal number of coherent regions and superpixels assignment to shape final segments. To solve our model, we propose a numerical algorithm based on the alternating direction method of multipliers (ADMM), whose iterations consist of two highly parallelizable sub-problems. We show each sub-problem enjoys closed-form solution which makes the ADMM iterations computationally very efficient. Extensive experiments on benchmark image segmentation datasets demonstrate that our proposed method in combination with an over-segmentation can provide high quality and competitive results compared to the existing state-of-the-art methods.

Theoretically, there are two kinds of models to calculate Locational Marginal Price (LMP): single-stage optimization model and multi-stage optimization model. In the paper, two mathematical models were analyzed and compared. Economic concept of LMP was explained and LMPs calculation was derived based on optimization methods. Characteristics of LMPs for different optimization models and how marginal generators and non-marginal generators can affect LMPs were discussed. In case study, the PJM five buses system was adopted to calculate LMPs under single-stage and multi-stage optimization model. The study results show that multi-stage model can accomplish economic dispatch more effective than single-stage model. But single-stage optimization model is better competent for LMP based settlement.

In this paper, we propose an approach for planar object recognition using binary local invariant feature and local histogram features, in scene image. First, the object is detected based on a homography which represents transformation of object in scene image from reference image, estimated from matching pairs of keypoints between two images. Then, local intensity histograms are computed from blocks inside scene object. In order to locate these blocks, a reference object is divided into many blocks then the corresponding blocks of the scene object are located based on the homography with assumption that object is a plane. To make the feature invariant to the illumination, the local histogram is computed from only intensity component (V) of HSV color of image. Similarity of two images is calculated from the similarity of their local histogram features and the matching ratio. The recognized object is the most similar reference object in data set. For evaluation, we experiment our method with a planar dataset from Stanford University which includes book cover, cd cover, painting, business card, etc. According to the result, our proposed method gets the higher accuracy compare to some other methods while remaining significant processing for an online application.

This paper proposes a bottom up approach for static hand gesture recognition. By extending the local orientation histogram feature, we make it applicable to the human hand, an object of very little texture. The key steps are augmenting the local orientation histogram feature vector with its relative image coordinates, and clustering the augmented vector to find a compact yet descriptive representation of the hand shape. The recognition result is given by collective voting of all the local orientation histogram features extracted from the hand region of the testing image. The matching score is evaluated by retrieval in the image feature database of the training hand gestures. Locality sensitive hashing is used to reduce the computational cost for retrieval. The experimental results on labelled real-world images demonstrate superior accuracy and efficiency of our algorithm.

Regularized soft K-means for discriminant analysis

Clustering is one of the most important unsupervised machine learning tasks. It is widely used to solve problems of intrusion detection, text analysis, image segmentation etc. Subspace clustering is the most important method for high-dimensional data clustering. In order to solve the problem of parallel subspace clustering for high-dimensional big data, this paper proposes a parallel subspace clustering algorithm based on spark named PSubCLUS which is inspired by SubCLU, a classical subspace clustering algorithm. While Spark is the most popular big data parallel processing platform currently, PSubCLUS uses the Resilient Distributed Datasets (RDD) provided by Spark to store data points in a distributed way. The two main performing stages of this algorithm, one-dimensional subspace clustering and iterative clustering, can be executed in parallel on each worker node of cluster. PSubCLUS also uses a repartition method based on the number of data points to achieve load balancing. Experimental results show that PSubCLUS has good parallel speedup and ideal load balancing effect, which is suitable for solving the parallel subspace clustering of high-dimensional big data.

Due to the numerous applications in computer vision, subspace clustering has been extensively studied in the past two decades. Most research puts emphasis on the spectral clustering-based methods in the recent years. This kind of methods usually extracts the affinity by the self-representation of the data points. Although they achieve the state-of-the-art results, the computation time will be unbearable when the number of the data points is large enough. In addition, the self-representation only considers the information provided by each single data point. In this paper, inspired by the idea of the superpixels in image segmentation, we first propose superpixels for subspace clustering with the large dataset. Then, we provide the strategy for the popular spectral clustering-based methods using these superpixels. Experimental results confirm that our superpixel-based subspace clustering methods can improve the computation speed dramatically. In addition, since the superpixels can consider the information provided by the group of data points, these methods can also improve the performance to some extent.

Subspace clustering has been widely used in image segmentation. These methods usually use superpixel segmentation to pre-segment image, while the superpixel segmentation method always divides image into superpixel blocks of similar shape and size, which results in poor segmentation and is time-consuming. In addition, the existing image segmentation methods based on subspace clustering don’t consider processing nonlinear structure data and complex noise. In order to solve the above problems, this paper proposes an image segmentation method (AMR_WT_NLMSC) based on non-convex low-rank multi-kernel clustering, which uses the adaptive morphological reconstruction seed segmentation (AMR_WT) for pre-segmentation, and designs non-convex low-rank multi-kernel subspace clustering (NLMSC) achieves the final segmentation. Experiments on real image datasets show that AMR_WT _NLMSC method has the more accurate segmentation effect.

Aiming at the difficult problem of coal gangue image segmentation in complex backgrounds, this paper proposes an image segmentation method based on the edge detection theory of the star algorithm. The pixel matrix is extracted one by one in the X and Y directions of the coal gangue image, and the central pixel of the matrix satisfying the monotonic change condition is assigned as 0. They are mapped to single-value images with equal size in turn, to realize the detection of coal and gangue edges in the images. The response strategy of adjusting matrix length n and assignment factor β in real-time by using the feedback result of the illuminance meter in changing illumination environment is given. Combining the edge detection method of the star algorithm with the morphological method, the fine segmentation of the coal gangue image is completed. The segmentation results are based on the segmentation results obtained by the AI algorithm, and the error rates of the pixel area and centroid coordinates of coal gangue are within 0.29%. This study provides a novel, precise and efficient solution to the problem of image edge detection and segmentation in complex backgrounds.

Problem statement: Segmentation plays an important role in medical imaging. Segmentation of an image is the division or separation of the image into dissimilar regions of similar attribute. In this study we proposed a methodology that integrates clustering algorithm and marker controlled watershed segmentation algorithm for medical image segmentation. The use of the conservative watershed algorithm for medical image analysis is pervasive because of its advantages, such as always being able to construct an entire division of the image. On the other hand, its disadvantages include over segmentation and sensitivity to false edges. Approach: In this study we proposed a methodology that integrates K-Means clustering with marker controlled watershed segmentation algorithm and integrates Fuzzy C-Means clustering with marker controlled watershed segmentation algorithm separately for medical image segmentation. The Clustering algorithms are unsupervised learning algorithms, while the marker controlled watershed segmentation algorithm makes use of automated thresholding on the gradient magnitude map and post-segmentation merging on the initial partitions to reduce the number of false edges and over-segmentation. Results: In this study, we compared K-means clustering and marker controlled watershed algorithm with Fuzzy C-means clustering and marker controlled watershed algorithm. And also we showed that our proposed method produced segmentation maps which gave fewer partitions than the segmentation maps produced by the conservative watershed algorithm. Conclusion: Integration of K-means clustering with marker controlled watershed algorithm gave better segmentation than integration of Fuzzy C-means clustering with marker controlled watershed algorithm. By reducing the amount of over segmentation, we obtained a segmentation map which is more diplomats of the several anatomies in the medical images.

A feature group weighting method for subspace clustering of high-dimensional data

Cluster analysis is an important technique of explorative data mining. It refers to a collection of statistical methods for learning the structure of data by solely exploring pairwise distances or similarities. Often meaningful structures are not detectable in these high-dimensional feature spaces. Relevant features can be obfuscated by noise from irrelevant measurements. These observations led to the design of subspace clustering algorithms, which can identify clusters that originate from different subsets of features. Hunting for clusters in arbitrary subspaces is intractable due to the infinite search space spanned by all feature combinations. In this work, we present a subspace clustering algorithm that can be applied for exhaustively screening all feature combinations of small- or medium-sized datasets (approximately 30 features). Based on a robustness analysis via subsampling we are able to identify a set of stable candidate subspace cluster solutions.