Honggang Zhang

Reference-Based Scheme Combined With K-SVD for Scene Image Categorization

A reference-based algorithm for scene image categorization is presented in this letter. In addition to using a reference-set for images representation, we also associate the reference-set with training data in sparse codes during the dictionary learning process. The reference-set is combined with the reconstruction error to form a unified objective function. The optimal solution is efficiently obtained using the K-SVD algorithm. After dictionaries are constructed, Locality-constrained Linear Coding (LLC) features of images are extracted. Then, we represent each image feature vector using the similarities between the image and the reference-set, leading to a significant reduction of the dimensionality in the feature space. Experimental results demonstrate that our method achieves outstanding performance.

Ordered histogram of shapemes: An ordered bag-of-features based shape descriptor for efficient shape matching

In this paper, we enhance the Shape Context-based descriptor, shapemes, by introducing an ordered bag-of-features model and dynamic programming. The proposed descriptor consists of a series of sub-histograms of shapemes, each of which represents a subset of sampled points. The division of the sampled points is based on their sequential positions on the contour of the shape, so the representation has intrinsic order and is therefore named ordered histogram of shapemes. Then dynamic programming is utilized for descriptor matching. The framework is effective and efficient owing to the following properties: 1) points division approach together with dynamic programming for invariance under the change of starting point, 2) Earth Movers Distance for discriminative power, and 3) pre-caculated shapemes dissimilarity matrix for fast descriptor distance calculation. Experiments on standard shape database and real world application scenario demonstrate the effectiveness and efficiency of the descriptor and the matching framework. We make our code and experimental data publicly available for future reference.

Multi-feature content-based product image retrieval based on region of main object

Content-based image retrieval (CBIR) has got an intense interest and seen considerable progress over the last decade. But most of the time it is only applied in laboratory. One important reason for this is the diversity of images. Different practical situations call for different taxonomy definitions of images, and lead to very different solutions. At present, and even in the foreseeable future, a general purpose CBIR system is not really possible. However, image search engines oriented at specific domains are feasible in technology and also have the actual demand. With the rapid development of electronic commerce, searching specific product by image has become one of the most attractive related research topics. In this paper, we propose a region-based method fit for the content-based retrieval of product images. The method focuses on two key issues: fast extraction of the main region, in which the product locates, as well as efficient shape and color features extraction. To show the validity of the proposed region-based method, compared experiments are carried out and illustrated on the PI 100 dataset.

Representative reference-set and betweenness centrality for scene image categorization

Reference-based image classification approach introduces a reference-set for both image representation and dictionary learning. It significantly reduces the dimensionality of represented images and shows outstanding performance even with randomly selected reference images and simple distance measure. In this paper, we improve upon existing work with two major contributions. First, we show that a more representative reference-set contributes to better classification accuracy. To this end, we carefully adapt the K-means clustering algorithm in the feature space to select a distinguished reference-set. Second, in the image classification process, we propose to represent each image by measuring its betweenness centrality in a social network composed of the representative reference-set in each class, leading to a more coherent distance measure that considers the overall connectivity between the probe image and the reference-set. Extensive experiment results demonstrate that our proposed scheme achieves better performance than existing methods.

Codebook optimization using word activation forces for scene categorization

Visual codebook based quantization of robust appearance descriptors extracted from local image patches is an effective means of capturing image statistics for texture analysis and natural scene classification. In this paper, based on the newly proposed statistics of word activation forces (WAFs), we optimize the codebook. Currently, codebooks are typically created from a set of training images using a clustering algorithm. However, these codebooks are often functionally limited due to redundancy. We show that WAFs can remove the redundancy efficiently. In the experiment, the proposed method achieved the state-of-the-art performance on the Caltech-101, fifteen natural scene categories and VOC2007 databases. The optimization method also offers insights into the success of several recently proposed images classification approaches, including vector quantization (VQ) coding in the Spatial Pyramid Matching (SPM), sparse coding SPM (ScSPM), and Locality-constrained Linear Coding (LLC).

A foreground segmentation method for mobile image retrieval system

Content-based image retrieval (CBIR) is an application of computer vision techniques to the image retrieval problem. That is, the problem of searching for digital images in large database. In this paper, we apply an image segmentation technique to an image retrieval system which is designed for the use on mobile devices. Given an image captured by the mobile devices, edge detection and region merging mechanisms are used in this segmentation technique to extract the ROI from a complex background scene. The proposed method automatically merges the regions that are initially segmented by mean shift segmentation, and then effectively extracts the object contour by the labeled regions as either background or foreground. With no users interaction, the experimental results show the method is more effective than other automatic segmentation methods.

Improving bag-of-words scheme for scene categorization

Abstract Bag-of-words (BoW) representation becomes one of the most popular methods for representing image content and has been successfully applied to object categorization. This paper uses the newly proposed statistics of word activation forces (WAFs) to reduce the redundancy in the codebook used in the BoW model. In such a way, the representation of image features is improved. In addition, the authors propose a method using soft inverse document frequency (Soft-IDF) to optimize BoW based image features. Given visual words and the dataset, each visual word appears in different amount of images and also different times in each particular image. Some of the visual words appear rare in contrary to the frequent ones. The proposed method balances this case. Experiments show encouraging results in scene categorization by the proposed approach.

Fusing multiple statistical features via explicit feature mapping for person re-identification

Person re-identification (Re-ID) across non-overlapping camera views is one of the challenging problems in surveillance video analysis. In this paper, we propose to combine multiple statistical features via explicit kernel feature mapping, and learn a linear metric model by local fisher discriminant analysis (LFDA) for person Re-ID. To strengthen the robustness of our representation, three complementary statistical characteristics, including histogram-like features, covariance matrix and expectation vector, were extracted from multiple spatial scales for each person image. Experimental results show that the proposed method works effectively on the popular benchmark data sets VIPeR and CUHK01 and yield impressive performance measured with Cumulative Match Characteristic curves (CMC).

MFSC: A new shape descriptor with robustness to deformations

In this paper, we propose a new shape descriptor, Multi-scale Fuzzy Shape Context (MFSC), highlighted by its robustness to deformations. A novel multi-scale fuzzy model is presented and applied on the widely used shape descriptor Shape Context to generate MFSC. The multi-scale fuzzy model can handle shape deformations of different scales, which makes MFSC robust to various deformations. Experiments on an articulated shape dataset demonstrate performance improvement gained by MFSC over existing methods. We also applied MFSC on a real-world application, Content-Based Product Image Retrieval, and the experimental results further validate its effectiveness. We make our code and experimental data publicly available for future reference.

A new multi-scale fuzzy model for Histogram-Based Descriptors

In this paper, we first propose a general Multi-Scale Fuzzy Model (MSFM) which handles distortions at different scales in Histogram-Based Descriptors (HBDs). This model can be applied both on one-dimensional HBDs and multi-dimensional HBDs. We then focus on applying MSFM on the widely used Shape Context for a Simplified Multi-scale Fuzzy Shape Context (SMFSC) descriptor. Fuzzy models are barely used in multi-dimensional HBDs due to the significant increase of computational complexity. We show that by introducing an intra-bin point location approximation and an approximate iterative fuzzification approach, the algorithm can be simplified and thus SMFSC hardly increases computational complexity. Experiments on standard shape dataset show that SMFSC improves upon the Inner Distance Shape Context. We also applied SMFSC on Content-Based Product Image Retrieval and the experimental results further validate the effectiveness of our model.