Yunxue Shao

Salient Local Binary Pattern for Ground-Based Cloud Classification

Ground-based cloud classification is challenging due to extreme variations in the appearance of clouds under different atmospheric conditions. Texture classification techniques have recently been introduced to deal with this issue. A novel texture descriptor, the salient local binary pattern (SLBP), is proposed for ground-based cloud classification. The SLBP takes advantage of the most frequently occurring patterns (the salient patterns) to capture descriptive information. This feature makes the SLBP robust to noise. Experimental results using ground-based cloud images demonstrate that the proposed method can achieve better results than current state-of-the-art methods.

Multiple Instance Learning Based Method for Similar Handwritten Chinese Characters Discrimination

This paper proposes a Multiple Instance Learning based method for similar handwritten Chinese characters discrimination. The similar handwritten Chinese characters recognition problem is first defined as a Multiple-instance learning problem. Then the problem is solved by the AdaBoost framework. The proposed method selects some self-adapting critical regions as weak classifiers, and therefore it is more suitable for the wide variability of writing styles. Our experimental results demonstrate that the proposed method outperforms the other state-of-the-art methods.

Text detection in natural images based on character classification

Text information in images is very important for image understanding. In this paper, a text location method based on character classification is proposed. The and-valley image (AVI) and the and-ridge image (ARI) are first extracted from the input image. Then character components are detected from the AVI and ARI respectively, and then these components are sent to a character classifier. Finally,text region can be generated by merging all the recognized components. This approach is robust to font style, font size, font color and the background complexity. It is demonstrated in the experiments that our method is efficient.

Illumination-invariant completed LTP descriptor for cloud classification

Cloud classification plays an essential role in a large number of applications. However, this issue is particularly challenging due to the extreme appearance variations under different atmospheric conditions. In this paper, a novel descriptor named illumination-invariant completed local ternary pattern (ICLTP) is proposed for cloud classification. The proposed descriptor effectively handles the illumination variations by introducing illumination invariant factor. Furthermore, the Quadratic-Chi metric, which is more suitable for comparing the difference between two histograms, is applied instead of Chi-Square metric. The experimental results demonstrate the superior performance of our strategy on two challenging cloud databases. Besides cloud classification, we further validate the proposed ICLTP operator on traditional texture classification, which show the good generalization ability.

Ground-based cloud classification using multiple random projections

Ground-based cloud classification plays an essential role in meteorological research and has received great concern in recent years. In this paper, a novel algorithm named multiple random projections (MRP) is proposed for ground-based cloud classification. The proposed algorithm uses an ensemble approach of MRP to obtain an optimized textons. Based on the textons, discriminative features can be obtained for classification. A series of experiments on two ground-based cloud databases (Kiel and IapCAS-E) are conducted to evaluate the efficiency of our proposed method. In addition, three current state-of-the-art methods, which include Patch, PCA, single random projection (SRP), are selected for comparison purpose. The experimental results show that our MRP method can achieved the best classification performance.

Modified Two-Class LDA Based Compound Distance for Similar Handwritten Chinese Characters Discrimination

This paper proposes a modified two-class LDA based compound distance for similar handwritten Chinese characters discrimination. First the definition of the Intersecting Subspace (IS) between two classes and the modified between-class scatter matrix is given. Then we prove that the modified between-class scatter matrix can supply additional information. Our experiments demonstrate that the additional information can be used to discriminate points in the IS and the proposed method outperforms the previous LDA based method.

Visual word density-based nonlinear shape normalization method for handwritten Chinese character recognition

In handwritten Chinese character recognition, the performance of a system is largely dependent on the character normalization method. In this paper, a visual word density-based nonlinear normalization method is proposed for handwritten Chinese character recognition. The underlying rationality is that the density for each image pixel should be determined by the visual word around this pixel. Visual vocabulary is used for mapping from a visual word to a density value. The mapping vocabulary is learned to maximize the ratio of the between-class variation and the within-class variation. Feature extraction is involved in the optimization stage, hence the proposed normalization method is beneficial for the following feature extraction. Furthermore, the proposed method can be applied to some other image classification problems in which scene character recognition is tried in this paper. Experimental results on one constrained handwriting database (CASIA) and one unconstrained handwriting database (CASIA-HWDB1.1) demonstrate that the proposed method outperforms the start-of-the-art methods. Experiments on scene character databases chars74k and ICDAR03-CH show that the proposed method is promising for some image classification problems.

Self-generation Voting Based Method for Handwritten Chinese Character Recognition

Abstract Voting strategy is very useful in pattern recognition. Many methods, like Boosting and Bagging, are proposed and are successfully used in some applications using this strategy. However, these methods are infeasible or unsuitable for handwritten Chinese character recognition because of the problems characteristics. In this paper, a self-generation voting method is proposed for further improving the recognition rate in handwritten Chinese character recognition. This method learns a set of parameters first for generating a set of samples from the test sample, and then classify these generated samples using a base-line classifier. At last, it gives the final recognition result by voting. Experimental results on two databases show that the proposed method is effective and useful in handwritten Chinese character recognition systems.

An efficient strategy for features combination

An important consideration in designing of visual object classification systems is the representation of images. In the past decades, significant improvement has been made, which can owe to two strategies: the proposal of new discriminative local image features together with the combination of existing local image features. In this paper, several strategies which provide the combination weights of different local features are enumerated, and then our own strategy, in which the weights of multiple features are determined by their contributions to the classification performance, is proposed. At the end of the paper, we provide our result in Visual Object Classes Challenge 2009 (VOC 2009), which shows that the strategy is competitive with state-of-the-art methods while the processing time is dramatically reduced.