Baochang Zhang

Land-use scene classification using multi-scale completed local binary patterns

In this paper, we introduce the completed local binary patterns (CLBP) operator for the first time on remote sensing land-use scene classification. To further improve the representation power of CLBP, we propose a multi-scale CLBP (MS-CLBP) descriptor to characterize the dominant texture features in multiple resolutions. Two different kinds of implementations of MS-CLBP equipped with the kernel-based extreme learning machine are investigated and compared in terms of classification accuracy and computational complexity. The proposed approach is extensively tested on the 21-class land-use dataset and the 19-class satellite scene dataset showing a consistent increase on performance when compared to the state of the arts.

Multi-Temporal Depth Motion Maps-Based Local Binary Patterns for 3-D Human Action Recognition

This paper presents a local spatio-temporal descriptor for action recognistion from depth video sequences, which is capable of distinguishing similar actions as well as coping with different speeds of actions. This descriptor is based on three processing stages. In the first stage, the shape and motion cues are captured from a weighted depth sequence by temporally overlapped depth segments, leading to three improved depth motion maps (DMMs) compared with the previously introduced DMMs. In the second stage, the improved DMMs are partitioned into dense patches, from which the local binary patterns histogram features are extracted to characterize local rotation invariant texture information. In the final stage, a Fisher kernel is used for generating a compact feature representation, which is then combined with a kernel-based extreme learning machine classifier. The developed solution is applied to five public domain data sets and is extensively evaluated. The results obtained demonstrate the effectiveness of this solution as compared with the existing approaches.

Gradient Local Auto-Correlations and Extreme Learning Machine for Depth-Based Activity Recognition

This paper presents a new method for human activity recognition using depth sequences. Each depth sequence is represented by three depth motion maps (DMMs) from three projection views (front, side and top) to capture motion cues. A feature extraction method utilizing spatial and orientational auto-correlations of image local gradients is introduced to extract features from DMMs. The gradient local auto-correlations (GLAC) method employs second order statistics (i.e., auto-correlations) to capture richer information from images than the histogram-based methods (e.g., histogram of oriented gradients) which use first order statistics (i.e., histograms). Based on the extreme learning machine, a fusion framework that incorporates feature-level fusion into decision-level fusion is proposed to effectively combine the GLAC features from DMMs. Experiments on the MSRAction3D and MSRGesture3D datasets demonstrate the effectiveness of the proposed activity recognition algorithm.

Boosting-Like Deep Convolutional Network for Pedestrian Detection

This paper proposes a boosting-like deep learning (BDL) framework for pedestrian detection. The fusion of handcrafted and deep learned features is considered to extract more effective representations. Due to overtraining on the limited training samples, over-fitting and convergence stability are two major problems of deep learning. We propose the boosting-like algorithm to enhance the system convergence stability through adjusting the updating rate according to the classification condition of samples in the training process. We theoretically give the derivation of our algorithm. Our approach achieves 15.85% and 3.81% reduction in the average miss rate compared with ACF and JointDeep on the largest Caltech dataset, respectively.

Hyperspectral image classification using Gradient Local Auto-Correlations

Spatial information has been verified to be helpful in hyperspectral image classification. In this paper, a spatial feature extraction method utilizing spatial and orientational auto-correlations of image local gradients is presented for hyperspectral imagery (HSI) classification. The Gradient Local Auto-Correlations (GLAC) method employs second order statistics (i.e., auto-correlations) to capture richer information from images than the histogram-based methods (e.g., Histogram of Oriented Gradients) which use first order statistics (i.e., histograms). The experiments carried out on two hyperspectral images proved the effectiveness of the proposed method compared to the state-of-the-art spatial feature extraction methods for HSI classification.

Adaptive Multi-class Correlation Filters

Correlation filters have attracted growing attention due to their high efficiency, which have been well studied for binary classification. However, by setting the desired output to be a fixed Gaussian function, the conventional multi-class classification based on correlation filters becomes problematic due to the under-fitting in many real-world applications. In this paper, we propose an adaptive multi-class correlation filters (AMCF) method based on an alternating direction method of multipliers (ADMM) framework. Within this framework, we introduce an adaptive output to alleviate the under-fitting problem in the ADMM iterations. By doing so, a closed-form sub-solution is obtained and further used to constrain the optimization objective, simplifying the entire inference mechanism. The proposed approach is successfully combined with the Histograms of Oriented Gradients (HOG) features, multi-channel features and convolution features, and achieves superior performances over state-of-the-arts in two multi-class classification tasks including handwritten digits recognition and RGBD-based action recognition.

Action recognition using completed local binary patterns and multiple-class boosting classifier

This paper, for the first time, introduces a multiple-class boosting scheme (MBS) to combine depth motion maps (DMMs) and completed local binary patterns (CLBP) for action recognition. DMMs derive from projecting depth frames onto three orthogonal Cartesian planes (front, side and top) and characterize the motion energy of an action, on which the CLBP features are further extracted. And then a new multi-class boosting method is used and leads to an effective decision-level classifier. Extensive experiments on the MSRAction3D and MSRGesture3D datasets indicate that the proposed MBS method achieves new state-of-the-art results.

GPCA-SIFT: A New Local Feature Descriptor for Scene Image Classification

In this paper, a new local feature descriptor called GPCA-SIFT is proposed for scene image classification. Like PCA-SIFT, we get the key points using the detection method in Scale Invariant Feature Transform (SIFT) and extract a 41 * 41 patch for each key point. Then we calculate the horizontal and vertical gradient of each pixel in the patch. However, instead of concatenating two gradient matrices, we directly work with the two-dimensional matrix and apply Generalized Principal Component Analysis (GPCA) to reduce it to a lower-dimensional matrix. Finally, we concatenate the reduced matrix and form a 1D vector. Compared with Principal Component Analysis (PCA), it preserves more spatial locality information. When applied in multi-class scene image classification, our proposed descriptor outperforms other related algorithms in terms of classification accuracy.

Multilinear Local Fisher Discriminant Analysis for Face Recognition

In this paper, a multilinear local fisher discriminant analysis (MLFDA) framework is introduced for tensor object dimensionality reduction and recognition. MLFDA achieves feature extraction by finding a multilinear projection to map the original tensor space into a tensor subspace that maximize the local between-class scatter as well as minimize the local within-class scatter. The experimental result shows that MLFDA has an outperformance.

Local Dual-Cross Ternary Pattern for Feature Representation

Extracting effective features is a fundamental issue in image representation and recognition. In this paper, we present a new feature representation method for image recognition based on Local Ternary Pattern and Dual-Cross Pattern, named Local Dual-Cross Ternary Pattern (LDCTP). LDCTP is a feature representation inspired by the sole textural structure of human faces. It is efficient and only quadruples the cost of computing Local Binary Pattern. Experiments show that LDCTP outperforms other descriptors.