Yuchou Chang

Unsupervised feature selection using clustering ensembles and population based incremental learning algorithm

This paper describes a novel feature selection algorithm for unsupervised clustering, that combines the clustering ensembles method and the population based incremental learning algorithm. The main idea of the proposed unsupervised feature selection algorithm is to search for a subset of all features such that the clustering algorithm trained on this feature subset can achieve the most similar clustering solution to the one obtained by an ensemble learning algorithm. In particular, a clustering solution is firstly achieved by a clustering ensembles method, then the population based incremental learning algorithm is adopted to find the feature subset that best fits the obtained clustering solution. One advantage of the proposed unsupervised feature selection algorithm is that it is dimensionality-unbiased. In addition, the proposed unsupervised feature selection algorithm leverages the consensus across multiple clustering solutions. Experimental results on several real data sets demonstrate that the proposed unsupervised feature selection algorithm is often able to obtain a better feature subset when compared with other existing unsupervised feature selection algorithms.

A facial expression recognition system based on supervised locally linear embedding

In this paper, a facial expression recognition system based on supervised locally linear embedding (SLLE) is introduced. The system consists of three modules: face detection, feature extraction with SLLE and classification. In face detection module, two independent characteristics, skin color characteristic and motion characteristic are used to detect face region, and a trained SVM is used to verify candidate regions. In feature extraction module, SLLE, a supervised learning algorithm that can compute low dimensional, neighborhood-preserving embeddings of high dimensional data is used to reduce data dimension and extract features. In classification module, minimum-distance classifier is used to recognize different expressions. The experiments show that the proposed method is superior to PCA-based method.

Sensitivity Encoding Reconstruction With Nonlocal Total Variation Regularization

In sensitivity encoding reconstruction, the issue of ill conditioning becomes serious and thus the signal‐to‐noise ratio becomes poor when a large acceleration factor is employed. Total variation (TV) regularization has been used to address this issue and shown to better preserve sharp edges than Tikhonov regularization but may cause blocky effect. In this article, we study nonlocal TV regularization for noise suppression in sensitivity encoding reconstruction. The nonlocal TV regularization method extends the conventional TV norm to a nonlocal version by introducing a weighted nonlocal gradient function calculated from the weighted difference between the target pixel and its generalized neighbors, where the weights incorporate the prior information of the image structure. The method not only inherits the edge‐preserving advantage of TV regularization but also overcomes the blocky effect. The experimental results from in vivo data show that nonlocal TV regularization is superior to the existing competing methods in preserving fine details and reducing noise and artifacts. Magn Reson Med, 2011.

Consensus unsupervised feature ranking from multiple views

Feature ranking is a kind of feature selection process which ranks the features based on their relevances and importance with respect to the problem. This topic has been well studied in supervised classification area. However, very few works are done for unsupervised clustering under the condition that labels of all instances are unknown beforehand. Thus, feature ranking for unsupervised clustering is a challenging task due to the absence of labels of instances for guiding the computations of the relevances of features. This paper explores the feature ranking approach within the unsupervised clustering area. We propose a novel consensus unsupervised feature ranking approach, termed as unsupervised feature ranking from multiple views (FRMV). The FRMV method firstly obtains multiple rankings of all features from different views of the same data set and then aggregates all the obtained feature rankings into a single consensus one. Experimental results on several real data sets demonstrate that FRMV is often able to identify a better feature ranking when compared with that obtained by a single feature ranking approach.

Nonlinear GRAPPA: A kernel approach to parallel MRI reconstruction

GRAPPA linearly combines the undersampled k‐space signals to estimate the missing k‐space signals where the coefficients are obtained by fitting to some auto‐calibration signals (ACS) sampled with Nyquist rate based on the shift‐invariant property. At high acceleration factors, GRAPPA reconstruction can suffer from a high level of noise even with a large number of auto‐calibration signals. In this work, we propose a nonlinear method to improve GRAPPA. The method is based on the so‐called kernel method which is widely used in machine learning. Specifically, the undersampled k‐space signals are mapped through a nonlinear transform to a high‐dimensional feature space, and then linearly combined to reconstruct the missing k‐space data. The linear combination coefficients are also obtained through fitting to the ACS data but in the new feature space. The procedure is equivalent to adding many virtual channels in reconstruction. A polynomial kernel with explicit mapping functions is investigated in this work. Experimental results using phantom and in vivo data demonstrate that the proposed nonlinear GRAPPA method can significantly improve the reconstruction quality over GRAPPA and its state‐of‐the‐art derivatives. Magn Reson Med, 2012.

Unsupervised video shot detection using clustering ensemble with a color global scale-invariant feature transform descriptor

Scale-invariant feature transform (SIFT) transforms a grayscale image into scale-invariant coordinates of local features that are invariant to image scale, rotation, and changing viewpoints. Because of its scale-invariant properties, SIFT has been successfully used for object recognition and content-based image retrieval. The biggest drawback of SIFT is that it uses only grayscale information and misses important visual information regarding color. In this paper, we present the development of a novel color feature extraction algorithm that addresses this problem, and we also propose a new clustering strategy using clustering ensembles for video shot detection. Based on Fibonacci lattice-quantization, we develop a novel color global scale-invariant feature transform (CGSIFT) for better description of color contents in video frames for video shot detection. CGSIFT first quantizes a color image, representing it with a small number of color indices, and then uses SIFT to extract features from the quantized color index image. We also develop a new space description method using small image regions to represent global color features as the second step of CGSIFT. Clustering ensembles focusing on knowledge reuse are then applied to obtain better clustering results than using single clustering methods for video shot detection. Evaluation of the proposed feature extraction algorithm and the new clustering strategy using clustering ensembles reveals very promising results for video shot detection.

Color-texture image segmentation by integrating directional operators into JSEG method

We presented a new color-texture image segmentation method that combined directional operators and the JSEG (J measure based SEGmentation) algorithm. In particular, two major contributions were set forth in this paper. (1) The two measures defined in JSEG and HSEG (H measure based SEGmentation) were discussed respectively and then we found the over-segmentation problem of JSEG could be attributed partly to the absence of color discontinuity in the J measure. Moreover, we have also proven in essence the H measure is isotropic operators for edge detection in images. (2) A new measure was proposed by integrating directional operators into the J measure, on which our segmentation method was based. The new segmentation method took account of both textural homogeneity and color discontinuity in local regions. Performance improvement due to the proposed modification was demonstrated on a variety of real color images.

Color quantization and image analysis for automated fruit quality evaluation

Machine vision has become an important non-destructive visual inspection technology for automation in the past two decades. Using machine vision for production automation can reduce operating costs and increase product value and quality. For agricultural products, color is often a good indicator of product quality and maturity. This paper presents a novel image-dependent color quantization technique designed specifically for real-time color evaluation in production automation applications. In contrast with more complex color space conversion techniques, the proposed method makes it easy for a human operator to specify and adjust color-preference settings for different color groups representing distinct quality or maturity levels. The performance of this robust color quantization and image analysis technique in evaluating fruit maturity and detecting skin delamination defects is demonstrated using Medjool date samples collected from field testing.

Matching book-spine images for library shelf-reading process automation

Machine vision has become an important visual inspection technology for many automation applications. Using machine vision for automation can reduce operating costs and increase efficiency and accuracy. This paper presents an image matching technique designed specifically for improving the library inventory (shelf-reading) process. In contrast with more complex color image matching techniques, the proposed method quantizes color images of book spines into a limited number of color indices and performs image matching on the quantized color index images. This approach simplifies and speeds up the processing and improves the overall inventory process. The potential performance of this robust color quantization and image matching technique is demonstrated by the results of preliminary experiments.

Dixon water‐fat separation in PROPELLER MRI acquired with two interleaved echoes

To propose a novel combination of robust Dixon fat suppression and motion insensitive PROPELLER (periodically rotated overlapping parallel lines with enhanced reconstruction) MRI.