小型无人机对地目标盘旋跟踪仿真

主要研究了小型无人机对地跟踪问题。针对小型无人机的盘旋状态,为了提高该情况下的目标形变,光影变化及遮挡等干扰的鲁棒性与实时性,采用基于抑制背景的MeanShift目标跟踪算法,即对目标周围和背景差异显著性较大的区域赋予较大的权值,从而把背景对目标的影响有效抑制,实现目标始终在跟踪方框内。通过架构模拟实验,该方案能够有效减小目标的连续形变以及光照变化对目标的影响;对于目标旁边景物的遮挡也有较强的抗干扰能力,而且该方案可靠性高、算法复杂度低、实时性较强。     

基于改进CamShift融合Kalman滤波的无人机目标跟踪研究

针对无人机目标跟踪过程中CamShift算法对目标颜色相似背景干扰和遮挡干扰鲁棒性差问题,对CamShift算法进行了改进。首先,针对CamShift算法模板信息单一,易受到颜色相似背景干扰的问题,提出基于H分量和LBP二维直方图模板的CamShift目标跟踪算法,改进算法提高了算法对相似目标干扰的鲁棒性,且有效帧率提高了约21%;针对目标跟踪过程中目标易受到障碍物遮挡的问题,在CamShift算法中引进了Kalman滤波预测机制,增强了跟踪算法在目标遮挡条件下的鲁棒性和跟踪效率,其中跟踪效率提高了约25%,每帧迭代所用时间下降了约36%。     

一种复杂背景下红外目标稳定跟踪算法

针对红外单目标在长期跟踪过程中的强背景干扰、遮挡、形变以及目标特征信息减弱等实际问题,提出了一种基于跟踪-学习-检测(Tracking-Learning-Detection,TLD)框架的红外目标稳定跟踪方法.该方法在压缩跟踪算法(Compressive Tracking,CT)的基础上替换广义的类Harr特征为HOG特征,引入互补随机测量矩阵,优化纹理和灰度特征信息的权重,同时引入卡尔曼滤波器记录空间上下文位置信息,以解决CT算法和TLD算法在目标被遮挡时的跟踪失效和全局检索问题.基于TLD算法框架和改进CT算法相结合的红外图像跟踪算法有效地解决了遮挡和强干扰问题,提升了算法的跟踪准确性和长期跟踪稳定性.实验结果表明,本文提出的算法在红外地面环境中能较好地实时稳定跟踪并保持良好的准确性和鲁棒性.     

一种改进的背景感知相关滤波跟踪算法

针对视觉目标跟踪复杂环境中的遮挡、运动模糊、形变以及背景干扰等问题,提出了一种改进的背景感知相关滤波跟踪算法,通过增加提取灰度特征(Histogram of Oriented Gradient,HOG)及颜色特征(Color Names,CN),提升了目标定位精度,并在此基础上采用多峰检测的模型更新策略抑制相似特征,解...     

一种改进的Camshift视频目标跟踪算法

鉴于连续自适应均值漂移(Camshift)算法在光照变化,相似背景颜色干扰及目标遮挡时鲁棒性不高,易造成跟踪错误等问题,提出了一种联合多特征和最大类间方差法的视频运动目标跟踪算法。该算法将色度直方图、梯度方向直方图和LBP纹理特征进行巧妙的融合,构建了一种高效的联合直方图目标外观特征模型,并在Camshift算法中嵌入最大类间方差法,增强目标和背景的区分度。不同场景的视频跟踪结果表明,改进算法有效克服了传统Camshift算法应对光照变化、颜色干扰和目标遮挡的缺点,与同类算法相比,具有更高的准确度和鲁棒性。     

基于改进Camshift算法的NAO机器人目标跟踪

针对Camshift算法应用于NAO机器人目标跟踪过程中,当目标受到相似颜色背景干扰或被物体遮挡时跟踪失败的问题,提出一种基于ORB特征检测和Kalman滤波多算法结合的目标跟踪方法。首先检测目标ORB特征点初始化搜索窗口,然后利用Kalman滤波作为目标运动状态的预测机制,以预测的位置初始化Camshift算法。利用Bhattacharyya距离判断跟踪窗口的收敛性,若受到背景干扰,则利用ORB算法对当前帧中的Kalman预测区域和目标模型进行特征点匹配,重新检测目标在视频帧中的位置。根据Kalman滤波预测目标被物体遮挡后可能的位置来更新预测器参数。实验结果表明,改进的算法能够在相似颜色背景干扰和目标遮挡的复杂环境下,连续稳定地跟踪运动目标。     

一种改进的长时间压缩感知跟踪算法

压缩感知跟踪(CT)算法具有简单、高效、实时的优点,但是却存在着跟踪窗口尺寸不能自适应变化,无法有效处理遮挡以及跟踪失败后的目标再发现等问题.为了解决上述问题,提出了一种改进的长时间压缩感知跟踪算法.所提出的算法采用多尺度的目标外观再匹配方法,使得跟踪窗口大小能够适应目标尺寸变化.此外,通过分析滑动窗口内跟踪窗口图像的整体特征变化来判定目标是否发生遮挡.为了解决跟踪器漂移问题,采用Haar特征在线生成检测器,实现目标的再发现.实验结果表明提出的算法相比原CT算法具有更好的鲁棒性和准确性.     

复杂场景下的加权粒子滤波行人跟踪方法

针对粒子滤波跟踪算法在行人目标遮挡、光线干扰以及背景与行人相似等情形下,目标易发生漂移、跟踪精度不高的问题,本文提出一种加权粒子滤波行人跟踪方法。该方法联合遮挡模型和Online Boosting算法,利用在线学习实时更新强分类器,并结合跟踪时建立的遮挡模型,以及行人运动时与上一次目标位置的距离、相似度等影响因子,对粒子权重进行重新构造,实现了复杂变化场景下的行人自适应跟踪。通过对PETS-L2S1公共数据集和自有数据集分别进行实验,可以得到本文提出的方法能有效去除目标遮挡、相似背景以及光线突变的干扰,实现稳定、准确、实时的行人跟踪。      

改进的Camshift与Kalman融合的目标跟踪算法

Camshift算法主要利用物体的颜色信息进行跟踪,在复杂背景条件下容易造成目标的跟丢,且在目标被遮挡时,也容易造成跟踪失效。本文提出了一种改进的Camshift目标跟踪算法。首先将目标图像的HSV模型的三个分量进行加权建立一种新的目标颜色模型,然后由对整帧图像计算反向投影改为比搜索窗口稍大的区域计算反向投影,减少了相似背景的干扰。同时为了解决遮挡问题,结合了Kalman滤波器,有效地预测了目标的位置。实验表明,本算法能够避免背景颜色干扰和解决遮挡问题,实现了对运动目标准确跟踪。     

复杂场景下基于改进DiMP算法的精确目标跟踪

针对DiMP目标跟踪算法在自然场景下遇到遮挡及背景干扰导致跟踪表现不佳的问题,提出了改进的DiMP精确目标跟踪算法。在图像预处理阶段创新性地设计了一个任意灰度块替换策略来丰富样本的信息;将特征提取网络ResNet-50提取的目标各阶段的特征图输入到设计的多尺度融合模块中进行正向和反向的充分融合,得到包含更多位置信息和语义信息的特征图;随后特征图输入到模板预测模块中进行在线更新操作,进而得到判别力更强的目标模板。实验表明：该算法在UAV123数据集的遮挡和背景干扰测试中的成功率和精确率分别提高8%、4.15%和9%、6.30%;同时,在VOT2018的EAO指标上提高1.36%,在UAV123的成功率和精确率指标上分别提高3.89%和3.06%。说明改进的DiMP算法在对遮挡与背景干扰问题上优势明显,进而提升了算法的整体表现。     

Robust Multi‐person Tracking for Real‐Time Intelligent Video Surveillance

We propose a novel multiple‐object tracking algorithm for real‐time intelligent video surveillance. We adopt particle filtering as our tracking framework. Background modeling and subtraction are used to generate a region of interest. A two‐step pedestrian detection is employed to reduce the computation time of the algorithm, and an iterative particle repropagation method is proposed to enhance its tracking accuracy. A matching score for greedy data association is proposed to assign the detection results of the two‐step pedestrian detector to trackers. Various experimental results demonstrate that the proposed algorithm tracks multiple objects accurately and precisely in real time.     

A new framework for particle detection in low-SNR fluorescence live-cell images and its application for improved particle tracking

Image denoising and signal enhancement are two common steps to improve particle contrast for detection in low-signal-to-noise ratio (SNR) fluorescence live-cell images. However, denoising may oversmooth features of interest, particularly weak features, leading to false negative detection. Here, we propose a robust framework for particle detection in which image denoising in the grayscale image is not needed, so avoiding image oversmoothing. A key to our approach is the new development of a particle enhancement filter based on the recently proposed particle probability image to obtain significantly enhanced particle features and greatly suppressed background in low-SNR and low-contrast environments. The new detection method is formed by combining foreground and background markers with watershed transform operating in both particle probability and grayscale spaces; dynamical switchings between the two spaces can optimally make use the information in images for accurate determination of particle position, size, and intensity. We further develop the interacting multiple mode filter for particle motion modeling and data association by incorporating the extra information obtained from our particle detector to enhance the efficiency of multiple particle tracking. We find that our methods lead to significant improvements in particle detection and tracking efficiency in fluorescence live-cell applications.     

子空间中的克隆选择人体运动分析方法

将基于视频的人体运动分析归结为高维空间中的优化问题,提出一种基于子空间构造和克隆选择的人体运动分析方法.该方法首先采用等距映射算法建立人体运动的低维流形空间,其次采用克隆选择算法在流形空间中进行姿态优化,最后通过引入时序信息提出序列克隆选择方法实现运动跟踪.该方法能准确分析视频中的人体运动,具有良好的计算效率和精度.     

An ensemble-based system for microaneurysm detection and diabetic retinopathy grading

Reliable microaneurysm detection in digital fundus images is still an open issue in medical image processing. We propose an ensemble-based framework to improve microaneurysm detection. Unlike the well-known approach of considering the output of multiple classifiers, we propose a combination of internal components of microaneurysm detectors, namely preprocessing methods and candidate extractors. We have evaluated our approach for microaneurysm detection in an online competition, where this algorithm is currently ranked as first, and also on two other databases. Since microaneurysm detection is decisive in diabetic retinopathy (DR) grading, we also tested the proposed method for this task on the publicly available Messidor database, where a promising AUC 0.90 ± 0.01 is achieved in a "DR/non-DR"-type classification based on the presence or absence of the microaneurysms.     

Object Modeling with Color Arrangement for Region‐Based Tracking

In this paper, we propose a new color histogram model for object tracking. The proposed model incorporates the color arrangement of the target that encodes the relative spatial distribution of the colors inside the object. Using the color arrangement, we can determine which color bin is more reliable for tracking. Based on the proposed color histogram model, we derive a mean shift framework using a modified Bhattacharyya distance. In addition, we present a method of updating an object scale and a target model to cope with changes in the target appearance. Unlike conventional mean shift based methods, our algorithm produces satisfactory results even when the object being tracked shares similar colors with the background.     

Multiple instance deep learning for weakly-supervised visual object tracking

Intelligently tracking objects with varied shapes, color, lighting conditions, and backgrounds is an extremely useful application in many HCI applications, such as human body motion capture, hand gesture recognition, and virtual reality (VR) games. However, accurately tracking different objects under uncontrolled environments is a tough challenge due to the possibly dynamic object parts, varied lighting conditions, and sophisticated backgrounds. In this work, we propose a novel semantically-aware object tracking framework, wherein the key is weakly-supervised learning paradigm that optimally transfers the video-level semantic tags into various regions. More specifically, give a set of training video clips, each of which is associated with multiple video-level semantic tags, we first propose a weakly-supervised learning algorithm to transfer the semantic tags into various video regions. The key is a MIL (Zhong et al., 2020) [1]-based manifold embedding algorithm that maps the entire video regions into a semantic space, wherein the video-level semantic tags are well encoded. Afterward, for each video region, we use the semantic feature combined with the appearance feature as its representation. We designed a multi-view learning algorithm to optimally fuse the above two types of features. Based on the fused feature, we learn a probabilistic Gaussian mixture model to predict the target probability of each candidate window, where the window with the maximal probability is output as the tracking result. Comprehensive comparative results on a challenging pedestrian tracking task as well as the human hand gesture recognition have demonstrated the effectiveness of our method. Moreover, visualized tracking results have shown that non-rigid objects with moderate occlusions can be well localized by our method.     

Moment-based multi-lane detection and tracking

Lane-detection methods are still facing robustness issues when confronted with challenging road surfaces, road markings and illumination conditions. Such combined challenges occur infrequently but are crucial for driving safety. Although advanced learning-based methods (using deep learning) demonstrate an impressive performance, they rely on plenty of training images for varying scenes and their performance is limited for scenes not covered by the training data. Also, multi-lane detection is indispensable for determining the exact position of both ego-car and surrounding vehicles as well as lane changing behavior on the road. In this paper we propose a new multi-lane detection algorithm, detecting all visible lane boundaries in front of the ego-car. In contrast to the Hough transforms often used for lane boundaries detection, our approach uses moments to calculate the deflection angles and the centroids of lane segments, achieving more precise lane boundaries. We propose a novel algorithm based on moments and Kalman filtering to achieve lane tracking. State-of-the-art neural-network-based methods are compared with the proposed method concretely. Experimental results show that our method outperforms other (recently published) multi-lane detection algorithms regarding detection rate as well as accuracy.     

Structural sparse coding seeds–active appearance model for object tracking

In this paper, we propose a tracking algorithm that can robustly handle appearance variations in tracking process. Our method is based on seeds–active appearance model, which is composed by structural sparse coding. In order to compensate for illumination changes, heavy occlusion and appearance self-updating problem, we proposed a mixture online learning scheme for modeling the target object appearance model. The proposed object tracking scheme involves three stages: training, detection and tracking. In the training stage, an incremental SVM model that directly measures the candidates samples and target difference. The proposed mixture generate–discriminative method can well separate two highly correlated positive candidates images. In the detection stage, the trained weighted vector is used to separate the target object in positive candidates images with respect to the seeds images. In the tracking stage, we employ the particle filter to track the object through an appearance adaptive updating algorithm with seeds–active constrained sparse representation. Based on a set of comprehensive experiments, our algorithm has demonstrated better performance than alternatives reported in the current literature.     

A support vector machine approach for detection of microcalcifications

In this paper, we investigate an approach based on support vector machines (SVMs) for detection of microcalcification (MC) clusters in digital mammograms, and propose a successive enhancement learning scheme for improved performance. SVM is a machine-learning method, based on the principle of structural risk minimization, which performs well when applied to data outside the training set. We formulate MC detection as a supervised-learning problem and apply SVM to develop the detection algorithm. We use the SVM to detect at each location in the image whether an MC is present or not. We tested the proposed method using a database of 76 clinical mammograms containing 1120 MCs. We use free-response receiver operating characteristic curves to evaluate detection performance, and compare the proposed algorithm with several existing methods. In our experiments, the proposed SVM framework outperformed all the other methods tested. In particular, a sensitivity as high as 94% was achieved by the SVM method at an error rate of one false-positive cluster per image. The ability of SVM to out perform several well-known methods developed for the widely studied problem of MC detection suggests that SVM is a promising technique for object detection in a medical imaging application.     

Object detection using multi-resolution mosaic in image sequences

Object detection in image sequences has a very important role in many applications such as surveillance systems, tracking and recognition systems, coding systems and so on. This paper proposes a unified framework for background subtraction, which is very popular algorithm for object detection in image sequences. And we propose an algorithm using spatio-temporal thresholding and truncated variable adaptation rate (TVAR) for object detection and background adaptation, respectively. Especially when the camera moves and zooms in on something to track the target, we generate multi-resolution mosaic which is made up of many background mosaics with different resolution, and use it for object detection. Some experimental results in various environments show that the averaged performance of the proposed algorithm is good.