Search

Low-resolution images are characterized by blurring, less texture information, and lack of detail. Visual object tracking for low-resolution imaging systems remains a challenging task. In this paper, we propose a discriminative correlation tracking algorithm based on a spatial attention mechanism for low-resolution imaging systems (LSDCT) to address these challenges. The key innovations of our proposed algorithm include adjustable windows and a spatial attention mechanism. We design a generic adjustable window to mitigate boundary effects and employ the spatial attention mechanism to highlight the target in low-resolution images. We conduct qualitative and quantitative evaluations on three well-known benchmark datasets: OTB100, TC128, and UAV123. Extensive experimental results indicate that the proposed approach is superior to state-of-the-art trackers.

A fast algorithm is proposed to tackle the constrained total generalised variation (TGV)-based image-restoration and reconstruction problems. The proposed algorithm proceeds by splitting: the non-smooth constrained TGV model is first decomposed into several sub-problems easier to solve, and then the linear gradient or proximity operators, including projections and shrinkages, of the sub-problems are individually called without inner iteration. The algorithm is highly parallel since most of its steps can be executed simultaneously. Image-restoration and reconstruction experiments demonstrate that the proposed algorithm outperforms several state-of-the-art TV-based methods both in accuracy and high-speed efficiency. Besides, the proposed method efficiently suppresses staircase effects and presents better visual impression.

To solve the homography estimation problem containing outliers and noise, a fast, robust, and accurate method is proposed. In this method, the outliers are rejected based on the differing characteristics of algebraic errors between outliers and inliers, and the homography is estimated by minimising the residual vector. The advantage of this method is in integrating the outlier rejection into the estimation pipeline. The computational complexity of the proposed method is not increased, and the random sample consensus algorithm is not needed to extract the inliers, as was previously necessary. Since the outlier rejection process is based on an algebraic criterion without computing the re-projection error at each step, the speed of the proposed method is improved. Several simulations based on synthetic and real images illustrate the performance of the proposed method in terms of subjective visual quality, objective quality measurement, and computational time. The experimental results demonstrate that the proposed method achieves accurate, efficient and robust homography estimation under different image transformation degrees and different outlier ratios.

Purpose The purpose of this paper is to improve the robustness of the traditional Bhattacharyya metric for the effect of histogram quantization in the histogram-based visual tracking. However, the traditional Bhattacharyya metric neglects the correlation of crossing-bin and is not robust for the effect of histogram quantization. Design/methodology/approach In this paper, the authors propose a visual tracking method via crossing-bin histogram Bhattacharyya similarity in the particle filter. Findings A crossing-bin matrix is introduced into the traditional Bhattacharyya similarity for measuring the reference histogram and the candidate histogram, and the basic tasks of measure such as maximum similarity of self and the triangle inequality are proven. The authors use the proposed measure in the particle filter visual tracking framework and address a model update strategy based on the crossing-bin histogram Bhattacharyya similarity to improve the robustness of visual tracking. Originality/value In the experiments using the famous challenging benchmark sequences, precision of the proposed method increases by 12.8 per cent comparing the traditional Bhattacharyya similarity and the cost time decreases by 38 times comparing the incremental Bhattacharyya similarity. The experimental results show that the proposed method can track the object robustly and rapidly under illumination change and occlusion.

Road markings are important information of transport systems for drivers or intelligent vehicle. Efficient road markings feature extraction is pre-requisite to road markings detection, recognition and visual localization. However, most of previous lane markings feature extractors are operating on conventional images, the feature extraction methods for omnidirectional images are rarely considered in road markings detection research. In this paper, three road markings feature extractors for omni-directional images are presented based on different characteristics of road markings. Those lane markings feature extractors are evaluated on real world panoramic image dataset. Experimental results show the feature extractor combing the edge and intensity distribution perform best.

In this paper, a fast algorithm is proposed to tackle the constrained total generalized variation (TGV) based image restoration problem. The proposed algorithm proceeds by splitting: the nonsmooth constrained TGV model is first decomposed into several subproblems easier to solve; then the linear gradient or proximity operators, including projections and shrinkages, of the subproblems are individually called without inner iteration. The algorithm is highly parallel since most of its steps can be executed simultaneously. Image restoration experiments demonstrate that the proposed algorithm outperforms several state-of-the-art methods both in speed and accuracy, while efficiently suppressing staircase effects and presenting better visual impression.

In view that the noise of image in monocular visual navigation, a Neigh-Shrink de-noising method based on Bayes-Shrink is proposed. First, the method uses Bayes-Shrink adaptive threshold replace common threshold as the contraction threshold of Neigh-Shrink, and then eliminate the high frequency sub-band of image obtained by monocular visual system after non-subsample Contourlet transform (NSCT) decomposition. Experiment results show that the improved Neigh-Shrink method in de-noising performance and visual effects have been improved, with better anti-noise performance and excellent edge protection, and can effectively be used in monocular visual navigation.