Dimensionality reduction of visual features for efficient retrieval and classification

Visual retrieval and classification are of growing importance for a number of applications, including surveillance, automotive, as well as web and mobile search. To facilitate these processes, features are often computed from images to extract discriminative aspects of the scene, such as structure, texture or color information. Ideally, these features would be robust to changes in perspective, illumination, and other transformations. This paper examines two approaches that employ dimensionality reduction for fast and accurate matching of visual features while also being bandwidth-efficient, scalable, and parallelizable. We focus on two classes of techniques to illustrate the benefits of dimensionality reduction in the context of various industrial applications. The first method is referred to as quantized embeddings, which generates a distance-preserving feature vector with low rate. The second method is a low-rank matrix factorization applied to a sequence of visual features, which exploits the temporal redundancy among feature vectors associated with each frame in a video. Both methods discussed in this paper are also universal in that they do not require prior assumptions about the statistical properties of the signals in the database or the query. Furthermore, they enable the system designer to navigate a rate versus performance trade-off similar to the rate-distortion trade-off in conventional compression.

Download Full-text

Fast and unique Tucker decompositions via multiway blind source separation

Bulletin of the Polish Academy of Sciences Technical Sciences ◽

10.2478/v10175-012-0051-4 ◽

2012 ◽

Vol 60 (3) ◽

pp. 389-405 ◽

Cited By ~ 8

Author(s):

G. Zhou ◽

A. Cichocki

Keyword(s):

Dimensionality Reduction ◽

Blind Source Separation ◽

Large Scale ◽

High Efficiency ◽

Source Separation ◽

Low Rank ◽

Large Scale Problems ◽

Rank Matrix ◽

Reduction Methods ◽

Low Rank Matrix

Abstract A multiway blind source separation (MBSS) method is developed to decompose large-scale tensor (multiway array) data. Benefitting from all kinds of well-established constrained low-rank matrix factorization methods, MBSS is quite flexible and able to extract unique and interpretable components with physical meaning. The multilinear structure of Tucker and the essential uniqueness of BSS methods allow MBSS to estimate each component matrix separately from an unfolding matrix in each mode. Consequently, alternating least squares (ALS) iterations, which are considered as the workhorse for tensor decompositions, can be avoided and various robust and efficient dimensionality reduction methods can be easily incorporated to pre-process the data, which makes MBSS extremely fast, especially for large-scale problems. Identification and uniqueness conditions are also discussed. Two practical issues dimensionality reduction and estimation of number of components are also addressed based on sparse and random fibers sampling. Extensive simulations confirmed the validity, flexibility, and high efficiency of the proposed method. We also demonstrated by simulations that the MBSS approach can successfully extract desired components while most existing algorithms may fail for ill-conditioned and large-scale problems.

Download Full-text

Automated Tire Visual Inspection Based on Low Rank Matrix Recovery

10.21203/rs.3.rs-109309/v1 ◽

2020 ◽

Author(s):

Guangxu Li ◽

Zhouzhou Zheng ◽

Yuyi Shao ◽

Jinyue Shen ◽

Yan Zhang

Keyword(s):

Visual Inspection ◽

Sparse Matrix ◽

Industrial Applications ◽

Convergence Speed ◽

Low Rank ◽

Inspection System ◽

Matrix Recovery ◽

Rank Matrix ◽

Low Rank Matrix Recovery ◽

Low Rank Matrix

Abstract Visual inspection is a challenging and widely employed process in industries. In this work, an automated tire visual inspection system is proposed based on low rank matrix recovery. Deep Network is employed to perform texture segmentation which benefits low rank decomposition in both quality and computational efficiency. We propose a dual optimization method to improve convergence speed and matrix sparsity by incorporating the improvement of the soft-threshold shrinkage operator by the weight matrix M. We investigated how incremental multiplier affects the decomposition accuracy and the convergence speed of the algorithm. On this basis, image blocks were decomposed into low-rank matrix and sparse matrix in which defects were separated. Comparative experiments have been performed on our dataset. Experimental results validate the theoretical analysis. The method is promising in false alarm, robustness and running time based on multi-core processor distributed computing. It can be extended to other real-time industrial applications.

Download Full-text