Cairong Zhao

Multi-scale gist feature manifold for building recognition

Multi-scale gist (MS-gist) feature manifold for building recognition is presented in the paper. It is described as a two-stage model. In the first stage, we extract the multi-scale gist features that represent the structural information of the building images. Since the MS-gist features are extrinsically high dimensional and intrinsically low dimensional, in the second stage, an enhanced fuzzy local maximal marginal embedding (EFLMME) algorithm is proposed to project MS-gist feature manifold to low-dimensional subspace. EFLMME aims to preserve local intra-class geometry and maximize local interclass margin separability of MS-gist feature manifold of different classes at the same time. To evaluate the performance of our proposed model, experiments were carried out on the Sheffield buildings database, compared with the existing works: (a) the visual gist based building recognition model (VGBR) and (b) the hierarchical building recognition model (HBR). Moreover, EFLMME is evaluated on Sheffield buildings database compared with some linear dimensionality reduction methods. The results show that the proposed model is superior to other models in practice of building recognition and can handle the building recognition problem caused by rotations, variant lighting conditions and occlusions very well.

Fuzzy local maximal marginal embedding for feature extraction

In graph-based linear dimensionality reduction algorithms, it is crucial to construct a neighbor graph that can correctly reflect the relationship between samples. This paper presents an improved algorithm called fuzzy local maximal marginal embedding (FLMME) for linear dimensionality reduction. Significantly differing from the existing graph-based algorithms is that two novel fuzzy gradual graphs are constructed in FLMME, which help to pull the near neighbor samples in same class nearer and nearer and repel the far neighbor samples of margin between different classes farther and farther when they are projected to feature subspace. Through the fuzzy gradual graphs, FLMME algorithm has lower sensitivities to the sample variations caused by varying illumination, expression, viewing conditions and shapes. The proposed FLMME algorithm is evaluated through experiments by using the WINE database, the Yale and ORL face image databases and the USPS handwriting digital databases. The results show that the FLMME outperforms PCA, LDA, LPP and local maximal marginal embedding.

Maximal local interclass embedding with application to face recognition

Dimensionality reduction of high dimensional data is involved in many problems in information processing. A new dimensionality reduction approach called maximal local interclass embedding (MLIE) is developed in this paper. MLIE can be viewed as a linear approach of a multimanifolds-based learning framework, in which the information of neighborhood is integrated with the local interclass relationships. In MLIE, the local interclass graph and the intrinsic graph are constructed to find a set of projections that maximize the local interclass scatter and the local intraclass compactness simultaneously. This characteristic makes MLIE more powerful than marginal Fisher analysis (MFA). MLIE maintains all the advantages of MFA. Moreover, the computational complexity of MLIE is less than that of MFA. The proposed algorithm is applied to face recognition. Experiments have been performed on the Yale, AR and ORL face image databases. The experimental results show that owing to the locally discriminating property, MLIE consistently outperforms up-to-date MFA, Smooth MFA, neighborhood preserving embedding and locality preserving projection in face recognition.

Local Maximal Marginal Embedding with Application to Face Recognition

Many problems in information processing involve some form of dimensionality reduction. This paper develops a new approach for dimensionality reduction of high dimensional data, called local maximal marginal (interclass) embedding (LMME), to manifold learning and pattern recognition. LMME can be seen as a linear approach of a multimanifolds-based learning framework which integrates the information of neighbor and class relations. LMME characterize the local maximal marginal scatter as well as the local intraclass compactness, seeking to find a projection that maximizes the local maximal margin and minimizes the local intraclass scatter. This characteristic makes LMME more powerful than the most up-to-data method, Marginal Fisher Analysis (MFA), and maintain all the advantages of MFA. The proposed algorithm is applied to face recognition and is examined using the Yale, AR, ORL and face image databases. The experimental results show LMME consistently outperforms PCA, LDA and MFA, owing to the locally discriminating nature. This demonstrates that LMME is an effective method for face recognition.

Minimum Standard Deviation Difference-Based Thresholding

Classic statistical thresholding methods do not consider special characteristic of practical images and fail to obtain satisfying segmentation results for some images with similar statistical distributions in the object and background. In this paper, a novel statistical thresholding method based on standard deviation difference is presented to solve this problem. The proposed method defines standard deviation difference as criterion for threshold selection, and determines the optimal threshold by minimizing it. The new method was compared with three conventional thresholding methods on a variety of infrared images and general real world images, and experimental results show its effectiveness.

An Image Thresholding Method Based on Human Visual Perception

A novel thresholding method based on human visual perception is presented in this paper. The method first utilizes statistical characteristics of an image to choose two gray levels as candidate thresholds by the properties of human visual perception, and then determines the one having minimum standard deviation sum as the optimal threshold. Choice of candidate thresholds reduces search space of thresholds and accelerates threshold selection. The proposed method was compared with several classic thresholding methods on a variety of nondestructive testing (NDT) images and general real world images, and experimental results show its effectiveness and efficiency.

Fuzzy maximal marginal embedding and its application

In this paper, we develops a new approach, called fuzzy maximal marginal embedding (FMME), combining LMME (local maximal marginal embedding) with fuzzy set theory, in which the fuzzy k-nearest neighbor (FKNN) is implemented to achieve the nature distribution information of original samples, and this information is utilized to redefine the affinity weights of neighborhood graph (intraclass and interclass ) instead of the weights of the binary pattern. We can reduce sensitivity of the method to substantial variations between samples caused by varying illumination and shape, viewing conditions. That makes FMME more powerful and robust than other method. The proposed algorithm is examined using Yale and ORL face image databases. The experimental results show FMME outperforms PCA, LDA, LPP and LMME.

Sparse Embedding Visual Attention Model

Visual attention (VA), defined as the ability of a biological or artificial visual system to rapidly detect potentially relevant parts of a visual scene, provides a general purpose solution for low level feature detection in a visual architecture. Numerous computational models of visual attention have been suggested during the last two decades. In saliency map of VA, how to select weights of each feature map that can correctly reflect the response salience between feature maps is important. A sparse embedding visual attention (SEVA) model, inspired by sparse representation, is presented. This paper describes the feature saliency index measured by sparse representation that adjusts the weights of each feature map in proportion of its average contribution to the saliency map. The proposed visual attention system is examined by using different scene images. Results show that the SEVA model consistently outperforms the traditional VA model, attributed to the adaptation of the weights of each feature map. Keyword: Visual attention, sparse representation, feature saliency index