Moving object extraction from unregistered image sequence

The problem of extracting a moving object from an unregistered image sequence can be reduced to a simple decision theory problem of detecting a signal in noise. The signal is depicted by the linear velocity of the moving object, and the noise is depicted by the 'vibration' experienced by the imaging sensor.<>     

Robust appearance-guided particle filter for object tracking with occlusion analysis

A major challenge for most tracking algorithms is how to address the changes of object appearance during tracking, incurred by large illumination, scale, pose variations and occlusions. Without any adaptability to these variations, the tracker may fail. In contrast, if adapts too fast, the appearance model is likely to absorb some improper part of the background or occluding objects. In this paper, we explore a tracking algorithm based on the robust appearance model which can account for slow or rapid changes of object appearance. Specifically, each pixel in appearance model is represented using mixture Gaussian models whose parameters are on-line learned by sequential kernel density approximation. The appearance model is then embedded into particle filter framework. In addition, an occlusion handling scheme is invoked to explicitly indicate outlier pixels and deal with occlusion events, thus avoiding the appearance model to be contaminated by undesirable outlier ‘thing’. Extensive experiments demonstrate that our appearance-based tracking algorithm can successfully track the object in the presence of dramatic appearance changes, cluttered background and even severe occlusions.     

基于图像遮挡分析的帧率上变换

帧率上变换也称作时域内插,主要应用于高端多媒体系统的帧率转换、低比特率视频系统的码率恢复以及无线视频通信系统的缺帧重构等领域.提出了一种基于图像遮挡分析的帧率上变换算法,实验结果表明,所提出的算法较当前已有的类似算法,插值图像的主观和客观质量均有所提高.     

Entropy-based metrics for the analysis of partial and total occlusion in video object tracking

Metrics measuring tracking reliability under occlusion that are based on mutual information and do not resort to ground truth data are proposed. Metrics for both the initialisation of the region to be tracked as well as for measuring the performance of the tracking algorithm are presented. The metrics variations may be interpreted as a quantitative estimate of changes in the tracking region due to occlusion, sudden movement or deformation of the tracked object. Performance metrics based on the Kullback-Leibler distance and normalised correlation were also added for comparison purposes. The proposed approach was tested on an object tracking scheme using multiple feature point correspondences. Experimental results have shown that mutual information can effectively characterise object appearance and reappearance in many computer vision applications.     

HOS-based image sequence noise removal.

In this paper, a new spatiotemporal filtering scheme is described for noise reduction in video sequences. For this purpose, the scheme processes each group of three consecutive sequence frames in two steps: 1) estimate motion between frames and 2) use motion vectors to get the final denoised current frame. A family of adaptive spatiotemporal L-filters is applied. A recursive implementation of these filters is used and compared with its nonrecursive counterpart. The motion trajectories are obtained recursively by a region-recursive estimation method. Both motion parameters and filter weights are computed by minimizing the kurtosis of error instead of mean squared error. Using the kurtosis in the algorithms adaptation is appropriate in the presence of mixed and impulsive noises. The filter performance is evaluated by considering different types of video sequences. Simulations show marked improvement in visual quality and SNRI measures cost as well as compared to those reported in literature.     

Adaptive integrated image segmentation and object recognition

The paper presents a general approach to image segmentation and object recognition that can adapt the image segmentation algorithm parameters to the changing environmental conditions. Segmentation parameters are represented by a team of generalized stochastic learning automata and learned using connectionist reinforcement learning techniques. The edge-border coincidence measure is first used as reinforcement for segmentation evaluation to reduce computational expenses associated with model matching during the early stage of adaptation. This measure alone, however, cannot reliably predict the outcome of object recognition. Therefore, it is used in conjunction with model matching where the matching confidence is used as a reinforcement signal to provide optimal segmentation evaluation in a closed-loop object recognition system. The adaptation alternates between global and local segmentation processes in order to achieve optimal recognition performance. Results are presented for both indoor and outdoor color images where the performance improvement over time is shown for both image segmentation and object recognition     

Urban object reconstruction using airborne laser elevation image and aerial image

Creating three-dimensional (3-D) models of real urban objects is an important goal in a wide variety of applications. This paper describes a method that utilizes airborne laser elevation images and aerial images for the 3-D reconstruction of urban objects. Our modeling approach uses the vertical geometric pattern analysis of elevation images. These patterns correspond to object contours and, thus, enable the extraction of the object. In addition, to provide realistic textured details, textures are cut from aerial images and mapped onto 3-D models. Our texture-mapping approach can avoid geometry mismatching and enable the automatic registration to determine the most reliable correspondence between projected outlines of 3-D models and contours of real objects shown in aerial images. Edge pairs, which are matched with projected outlines, are detected from aerial images. In order to minimize mismatching, we apply the voting technique based on the generalized Hough transform. Experimental results show that 3-D reconstruction of urban objects is generally successful.     

Contour image sequence compression through motion analysis and hybrid coding method

This paper presents a motion analysis algorithm (MAA) and a hybrid coding method for contour image sequence compression. The contour image sequence consists of objects moving and rotating in a 3-D world with occlusion, shape, and size variations from frame to frame. The MAA separates the moving image sequence into several object-oriented subsequences (OOSs). In each OOS, the component is either stationary or moves smoothly, and the motion parameters can be easily estimated. The first and last frames of OOS are key frames, and the others are in-between frames. The key frames are unpredictable, and the entire frames need to be encoded. The in-between frames are compensable, and they are encoded by the motion parameter coding. The hybrid coder uses vectorgraph coding to remove spatial redundancy of the key frames and motion parameter coding to reduce the temporal redundancy of the OOSs. The motion parameters are encoded as combinations of 2-D translation, 2-D rotation, and scaling. There are many applications for contour image sequence compression. The cartoon image sequence (a sequence of line drawing sketches) and the high-frame-rate videophone for sign language transmission are good examples. Experiments show that our method encodes the contour image sequence at a very high compression ratio without losing intelligibility.     

基于图像视频序列分析的篮球飞行轨迹跟踪

对篮球飞行轨迹进行图像视频跟踪,可以精准分析篮球的飞行轨迹曲线,从而指导篮球投篮训练。提出基于图像视频序列分析的篮球飞行轨迹跟踪方法。进行篮球飞行轨迹的图像视频序列采集,采用Radon尺度变换方法进行图像的降噪处理,拟合跟踪图像视频序列的灰度像素值特征点,实现飞行轨迹的图像量化分析。最后进行仿真测试,结果表明,采用该方法进行篮球飞行轨迹跟踪的拟合度较高,特征点提取的精度较高,对指导篮球训练具有积极意义。     

Region filling and object removal by exemplar-based image inpainting

A new algorithm is proposed for removing large objects from digital images. The challenge is to fill in the hole that is left behind in a visually plausible way. In the past, this problem has been addressed by two classes of algorithms: 1) "texture synthesis" algorithms for generating large image regions from sample textures and 2) "inpainting" techniques for filling in small image gaps. The former has been demonstrated for "textures"--repeating two-dimensional patterns with some stochasticity; the latter focus on linear "structures" which can be thought of as one-dimensional patterns, such as lines and object contours. This paper presents a novel and efficient algorithm that combines the advantages of these two approaches. We first note that exemplar-based texture synthesis contains the essential process required to replicate both texture and structure; the success of structure propagation, however, is highly dependent on the order in which the filling proceeds. We propose a best-first algorithm in which the confidence in the synthesized pixel values is propagated in a manner similar to the propagation of information in inpainting. The actual color values are computed using exemplar-based synthesis. In this paper, the simultaneous propagation of texture and structure information is achieved by a single, efficient algorithm. Computational efficiency is achieved by a block-based sampling process. A number of examples on real and synthetic images demonstrate the effectiveness of our algorithm in removing large occluding objects, as well as thin scratches. Robustness with respect to the shape of the manually selected target region is also demonstrated. Our results compare favorably to those obtained by existing techniques.     

Unsupervised image segmentation based on analysis of binary partition tree for salient object extraction

This paper proposes an unsupervised image segmentation approach aimed at salient object extraction. Starting from an over-segmentation result of a color image, region merging is performed using a novel dissimilarity measure considering the impact of color difference, area factor and adjacency degree, and a binary partition tree (BPT) is generated to record the whole merging sequence. Then based on a systematic analysis of the evaluated BPT, an appropriate subset of nodes is selected from the BPT to represent a meaningful segmentation result with a small number of segmented regions. Experimental results demonstrate that the proposed approach can obtain a better segmentation performance from the perspective of salient object extraction.     

Space-time decomposition: exploiting the full information of atraining sequence for an adaptive array

The authors describe a space-time decomposition (STD) approach for adjusting the array coefficients of a space-time array, which is inherently parallel and gives higher output signal-to-noise ratios than conventional approaches. The length of the training sequence limits only the number of antenna elements in an adaptive array. The length of the tapped delay lines can be adjusted independently of the length of the training sequence     

Fan filters, the 3-D Radon transform, and image sequence analysis

This paper develops a theory for the application of fan filters to moving objects. In contrast to previous treatments of the subject based on the 3-D Fourier transform, simplicity and insight are achieved by using the 3-D Radon transform. With this point of view, the Radon transform decomposes the image sequence into a set of plane waves that are parameterized by a two-component slowness vector. Fan filtering is equivalent to a multiplication in the Radon transform domain by a slowness response function, followed by an inverse Radon transform. The plane wave representation of a moving object involves only a restricted set of slownesses such that the inner product of the plane wave slowness vector and the moving object velocity vector is equal to one. All of the complexity in the application of fan filters to image sequences results from the velocity-slowness mapping not being one-to-one; therefore, the filter response cannot be independently specified at all velocities. A key contribution of this paper is to elucidate both the power and the limitations of fan filtering in this new application. A potential application of 3-D fan filters is in the detection of moving targets in clutter and noise. For example, an appropriately designed fan filter can reject perfectly all moving objects whose speed, irrespective of heading, is less than a specified cut-off speed, with only minor attenuation of significantly faster objects. A simple geometric construction determines the response of the filter for speeds greater than the cut-off speed.     

Boosting image object retrieval and indexing by automatically discovered pseudo-objects

State-of-the-art object retrieval systems are mostly based on the bag-of-visual-words representation which encodes local appearance information of an image in a feature vector. An image object search is performed by comparing query object’s feature vector with those for database images. However, a database image vector generally carries mixed information of the entire image which may contain multiple objects and background. Search quality is degraded by such noisy (or diluted) feature vectors. To tackle this problem, we propose a novel representation, pseudo-objects – a subset of proximate feature points with its own feature vector to represent a local area, to approximate candidate objects in database images. In this paper, we investigate effective methods (e.g., grid, G-means, and GMM–BIC) to estimate pseudo-objects. Additionally, we also confirm that the pseudo-objects can significantly benefit inverted-file indexing both in accuracy and efficiency. Experimenting over two consumer photo benchmarks, we demonstrate that the proposed method significantly outperforms other state-of-the-art object retrieval and indexing algorithms.     

薄云薄雾影响下目标影像自适应增强

卫星图像在拍摄过程中,由于云雾等因素影响,导致影像的亮度增加,目标影像对比度和分辨率下降。针对薄云薄雾对卫星图像的影响,提出目标影像的自适应增强算法。实验表明,该算法具有较强的自适应性。提高了薄云薄雾影响下的目标影像的对比度,恢复了该区域的亮度,同时,保持了薄云薄雾影响弱的区域的对比度和亮度,达到了目标影像自适应增强的目的。     

基于小波图像金字塔的工件目标检测与定位研究

在工业生产线上的产品加工或质量视觉检测的过程中,经常要完成待加工工件的目标图像匹配或产品的视觉定位,为此,提出一种基于小波图像金字塔的相关匹配方法。基于特征的匹配方法如SIFT和HOG特征方法,虽然有很好的匹配精度和稳健性,但其计算量相对较大,通常不易满足现场实时应用需求。而提出的基于小波图像金字塔的工件目标检测与定位方法,先从低分辨率图像进行粗匹配,再在高分辨率图像上进行精确匹配,从而在基本不降低匹配精度的前提下,大大提高了匹配速度。多图像的目标检测实验以及在点胶机上的应用实践均表明文中所提算法优于传统基于模型的目标定位和识别方法,能满足更多的现场实时应用要求。     

Extraction of fluorescent dot traces from a scanning laser ophthalmoscope image sequence by spatio-temporal image analysis: Gabor filter and radon transform filtering

The scanning laser ophthalmoscope (SLO) allows the tracking of fluorescent dot motion, thereby enabling the flow velocities in perimacular capillaries to be directly measured. These can serve as an important index of local retinal soundness or reflect the whole body circulation status in disorders such as diabetes. Although it is possible to perceive moving fluorescent dots with the human eye, they are so faint and unstable that it is difficult to detect them by conventional digital still-image processing methods. To solve this problem, we generated spatio-temporal images of the fluorescent dots in a capillary and applied Gabor filters tuned to the direction of the traces in order to detect them. Finally, by discriminating and integrating the output using two levels of threshold, we were able to extract their traces. Because the medium-size Gabor filter requires a considerable amount of time for two-dimensional convolution calculation, we prove that there is a certain equivalence between the Gabor filter, the radon transform, and the Hough transform. In the light of this, we propose a form of radon transform filtering that includes a radon transform Gabor filter as a very long Gabor filter. This allows a whole trace to be detected in a single step with a one-dimensional convolution, thereby shortening the processing time. In an experiment, 60% of the traces could be detected without error, which is sufficient to allow the mean flow velocity in a capillary to be measured.