Locality-constrained Linear Coding Algorithm Research Articles

A Dual Locality-Constrained Linear Coding Algorithm for Image Classification

The locality-constrained linear coding (LLC) algorithm has been developed as an effective means for image classification. This algorithm uses locality linear constraints to encode feature points of the image, achieving higher classification accuracy than that of Spatial Pyramid Matching (SPM), spatial pyramid matching using sparse coding, and other traditional algorithms. However, the LLC algorithm uses only the locality information for the visual words in the dictionary while rarely using mutual information between the neighbouring feature points, causing severe ambiguities. Based on the principle of the locality correlation of images, we propose a new algorithm, named the dual locality-constrained linear coding algorithm (DLLC), which applies the locality information of the feature points and visual words and uses the discriminant information provided by the nearest neighbouring feature points. The experimental results demonstrate that the accuracy of the DLLC algorithm is higher than that of the LLC algorithm, particularly when the image category is large, but the set of dictionary training data is small.

Kernelized temporal locality learning for real-time visual tracking

Linear representation-based methods play an important role in the development of the target appearance modeling in visual tracking. However, such linear representation scheme cannot accurately depict the nonlinearly distributed appearance variations of the target, which often leads to unreliable tracking results. To fix this issue, we introduce the kernel method into the locality-constrained linear coding algorithm to comprehensively exploit its nonlinear representation ability. Further, to fully consider the temporal correlation between neighboring frames, we develop a point-to-set distance metric with L2, 1 norm as the temporal smoothness constraint, which aims to guarantee that the object between the two consecutive frames should be represented by the similar dictionaries temporally. Experimental results on Object Tracking Benchmark show that the proposed tracker achieves promising performance compared with other state-of-the-art methods.

Bearing Fault Diagnosis Based on Improved Locality-Constrained Linear Coding and Adaptive PSO-Optimized SVM

A novel bearing fault diagnosis method based on improved locality-constrained linear coding (LLC) and adaptive PSO-optimized support vector machine (SVM) is proposed. In traditional LLC, each feature is encoded by using a fixed number of bases without considering the distribution of the features and the weight of the bases. To address these problems, an improved LLC algorithm based on adaptive and weighted bases is proposed. Firstly, preliminary features are obtained by wavelet packet node energy. Then, dictionary learning with class-wise K-SVD algorithm is implemented. Subsequently, based on the learned dictionary the LLC codes can be solved using the improved LLC algorithm. Finally, SVM optimized by adaptive particle swarm optimization (PSO) is utilized to classify the discriminative LLC codes and thus bearing fault diagnosis is realized. In the dictionary leaning stage, other methods such as selecting the samples themselves as dictionary and K-means are also conducted for comparison. The experiment results show that the LLC codes can effectively extract the bearing fault characteristics and the improved LLC outperforms traditional LLC. The dictionary learned by class-wise K-SVD achieves the best performance. Additionally, adaptive PSO-optimized SVM can greatly enhance the classification accuracy comparing with SVM using default parameters and linear SVM.

Action Recognition Using Weighted Locality-Constrained Linear Coding

Recently, locality-constrained linear coding (LLC) as a coding strategy has attracted much attention, due to its better reconstruction than sparse coding and vector quantization. However, LLC ignores the weight information of codewords during the coding stage, and assumes that every selected base has same credibility, even if their weights are different. To further improve the discriminative power of LLC code, we propose a weighted LLC algorithm that considers the codeword weight information. Experiments on the KTH and UCF datasets show that the recognition system based on WLLC achieves better performance than that based on the classical LLC and VQ, and outperforms the recent classical systems.

Salient object detection via global and local cues

Previous saliency detection algorithms used to focus on low level features directly or utilize a bunch of sample images and manually labeled ground truth to train a high level learning model. In this paper, we propose a novel coding-based saliency measure by exploring both global and local cues for saliency computation. Firstly, we construct a bottom-up saliency map by considering global contrast information via low level features. Secondly, by using a locality-constrained linear coding algorithm, a top-down saliency map is formulated based on the reconstruction error. To better exploit the local and global information, we integrate the bottom-up and top-down maps as the final saliency map. Extensive experimental results on three large benchmark datasets demonstrate that the proposed approach outperforms 22 state-of-the-art methods in terms of three popular evaluation measures, i.e., the Precision and Recall curve, Area Under ROC Curve and F-measure value. Furthermore, the proposed coding-based method can be easily applied in other methods for significant improvement.