Search

Supplement 1 (Tunable bilayer dielectric metasurface via stacking magnetic mirrors: supplemental document)

In this paper, a segmented minimum noise fraction (MNF) transformation is proposed for efficient feature extraction of hyperspectral images (HSIs). The original bands can be partitioned into several highly correlated subgroups based on the correlation matrix image of the hyperspectral data. The MNF is implemented separately on each subgroup of the data, and then, the Bhattacharyya distance is used as the band separability measure for feature extraction. Consequently, the extracted features can then be significantly classified using state-of-art classifiers, i.e., k-NN or SVM. Experiments on two benchmark HSIs collected by AVIRIS and ROSIS demonstrate that the proposed method significantly reduces the transformation time in comparison with the conventional MNF. The Fisher scalars' criterion shows that the class separability with the segmented MNF is the best, and the extracted features even exhibit higher classification accuracy compared with the PCA or MNF. HighlightsA segmented minimum noise fraction (MNF) transformation is proposed for efficient feature extraction of hyperspectral images (HSIs).The proposed method significantly reduces the transformation time in comparison with the conventional MNF.The class separability of the extracted features is improved.The extracted features by SMNF even exhibit higher classification accuracy compared with the PCA or MNF.

A novel optical flow estimation is proposed in this paper, which addresses some issues including the credible estimation of optical flow and the prevention of over-smoothing across motion boundaries. Our main contribution is that we estimated the optical flow by a nonlinear filtering process instead of an energy minimizing process, the latter often needs the corresponding smoothing constraint to be restricted to some form of convex and differentiable entity. In this way, it avoids some restrictions needed by the regularization. So we can choose a nonlinear filter from more flexible forms, which can help to deal with flow discontinuities more efficiently in some sense. We modified and extended a scalar 2D bilateral filter to optical flow field as the wanted nonlinear filter. Qualitative and quantitative results show that the new method can produce a reliable result.

This paper proposes a fast and stable automatic mosaic method of the unmanned aerial vehicle (UAV) images based on the Kalman filter. Firstly, the features of the unmanned aerial vehicle Images are analyzed. Then, a Kalman filter was proposed for predicting the search area of feature points after analyzing the movement model of the overlap areas in the images. The Kalman filter helps to find the useful feature points in the specific areas within a short time. Following the analysis result, the detail steps of the method are finally presented. The experimental results show that the proposed method not only ensure the successful execution of automatic mosaic for the UAV video images, but also can reduce the time-cost.

Key-frames are representative frames in a shot. Key frame extraction is one of the basic procedures relating to video retrieval and indexing. In view of the surveillance video characteristic and the user attention focus, this paper proposed a key frame extraction method based on multi-scale phase-based local features. Prior to key frame extraction, the video should be segmented into shots. Then, find the interest points in the head of the moving target extracted by adaptive background mixture Gaussian models, mark the candidate key frame which has a certain number of interest points matching with the given target model. Lastly, for each shot, extract key frame which has the best similar match. Experimental results demonstrate that the proposed method is feasible and effective.

Elastic image registration plays an important role in medical image registration. For elastic image registration based on landmarks of sub-images, optimization algorithm is applied to extract landmarks. But local maxima of similarity measure make optimization difficult to convergence to global maximum. The registration error will lead to location error of landmarks and lead to unexpected elastic transformation results. In this paper, an elastic image registration method using attractive and repulsive particle swarm optimization (ARPSO) is proposed. For each subimage, rigid registration is done using ARPSO. In attractive phase, particles converge to promise regions in the search space. In repulsive phase, particles are repelled each other along opposition directions and new particles are created, which might avoid premature greatly. Next, thin plate spline transformation is used for the elastic interpolation between landmarks. Experiments show that our method does well in the elastic image registration experiments.

Human detection is important in visual surveillance systems. Especially, counting the number of moving objects and locating the head of an object is a difficult problem in human detection. In this paper, a novel method is used to count and locate moving objects in thermal images, which uses thermal and position information to compute the fuzzy weighted projection histogram. The peaks of the fuzzy weighted projection histogram could be sharper and more separated. In the meantime, a potential function clustering method is used to count the number and locate the position of peaks in the histogram, which is the number and the head position of moving objects. Experiments show that our method is effective and feasible to count and locate moving objects in thermal images.

In this paper, an online hand-drawn graphic symbol recognition algorithm based on hidden Markov models is presented. A rearrangement strategy is applied to the hand-drawn symbol points in order to alleviate the influence of the difference in drawing sequence. Based on rearranged drawing points, global distance measure and local angle feature are extracted as the feature vector. After the quantization, a discrete HMM is used as the core recognizer. The experiment shows the recognition rate of our system can be above 85%.

A new effective soft clustering method-sectional set fuzzy C-means (S2FCM) clustering is presented, in which the ideas of traditional "hard" clustering (HCM) and "fuzzy" clustering (FCM) are united and generalized. A effective /spl lambda/ value (/spl lambda//spl isin/~0,1\ is a factor of fuzzy sectional set) of S2FCM is /spl lambda/=0.5+1/2c (c is the total number of clusters). The main idea of the proposed method accords with procedure of a human's partition process in pattern recognition. Experimental results show S2FCM clustering method's effectiveness and advantage.

A new clustering method-the potential function partial weighted fuzzy C-mean (PWFCM) clustering method is presented. In this method samples' typicality is decided by a potential distribution function in sample feature space. According to a sample set, the problem of how to adaptively determine the weighted matrix W is also discussed. Because the different influences on the partition by different samples in the feature space is considered, the proposed method can effectively partition perplexing data sets. Finally, the results of two experiments are compared with satisfactory results.

A method of detecting firing particles in sequence images is proposed in this paper. The features of the firing particles in sequence images are analyzed and described in fuzzy linguistic variable. Adaptive fuzzy local thresholds are determinated to detect the moving objects. Image information fusion techniques are applied to reduce noise particles also. This method is feasible and satisfied with experiment results.

A method of detecting and tracking firing particle flow in sequence images is proposed in this paper. The features of the firing particle flow in sequence images are analyzed and described using a fuzzy linguistic variable that is used to detect the moving objects. Moving trajectory tracking of firing particle flow is accomplished adaptively also. This method is feasible and satisfied with experiment results.