Recursive bayesian estimation using gaussian sums

The Bayesian recursion relations which describe the behavior of the a posteriori probability density function of the state of a time-discrete stochastic system conditioned on available measurement data cannot generally be solved in closed-form when the system is either non-linear or nongaussian. In this paper a density approximation involving convex combinations of gaussian density functions is introduced and proposed as a meaningful way of circumventing the difficulties encountered in evaluating these relations and in using the resulting densities to determine specific estimation policies. It is seen that as the number of terms in the gaussian sum increases without bound, the approximation converges uniformly to any density function in a large class. Further, any finite sum is itself a valid density function unlike many other approximations that have been investigated.     

Facial model estimation from stereo/mono image sequence

Facial model coding is an integral part in MPEG-4 related applications. The generation of the facial model usually requires stereoscopic view of the face in the pre-processing stage. Although facial model can be successfully estimated from two stereo facial images, the occlusion effect and imprecise location of the feature point prohibit obtaining an accurate facial model. In this paper, several facial model estimation (FME) algorithms are proposed in order to find the precise facial model from a stereo or mono image sequence. Since a sequence of images is used to find the facial model, the problem of occlusion effects is less serious. An accurate facial model (within 7.21% error) can still be obtained by our schemes, even without the prior information on the three-dimensional position of the head with respect to the camera and the rotation axis/angle of the head's movement. This is the largest error of all FME algorithms presented in this paper when the subject does not wear eyeglasses. In addition, our schemes do not require precise camera parameters and avoid tedious camera calibration, thereby, simplifying the facial model extraction.     

Reconstruction of a discontinuous function from a few fourier coefficients using bayesian estimation

The goal of this paper is the application of spectral methods to the numerical solution of conservation law equations. Spectral methods furnish estimates of the firstn Fourier coefficients of the solution. But since the solutions of conservation law equations can have discontinuities, the estimate of the solution by summing the firstn terms of the Fourier series will haveO(1/n) error, even if the Fourier coefficients are known to high accuracy. But if the solution could be accurately reconstructed from its Fourier coefficients, spectral methods could be used effectively in these problems. A method for doing this is to assume a probability distribution for functions. Functions which are smooth away from the discontinuity are assumed to be likely, and those which are not smooth away from the discontinuity are assumed to be unlikely. Then a reconstruction algorithm is chosen by minimizing the expected error over all algorithms. It is possible to put the smoothness assumptions mentioned earlier into an infinite-dimensional Gaussian probability distribution, and then the minimum-error algorithm is well-known and fairly simple to construct and apply. If the Fourier coefficients of the reconstructed function are known exactly, then this approach gives very good results. But when used with Fourier coefficients obtained from a spectral approximation to Burgers' equation, the results were much less impressive, probably because the coefficients were not known very accurately. It is possible to construct filters that reconstruct a function using Legendre or Chebyshev coefficients for information instead Fourier coefficients. It is found that the performance of these filters is similar to the Fourier case.     

Integrated position estimation using aerial image sequences

Presents an integrated system for navigation parameter estimation using sequential aerial images, where the navigation parameters represent the positional and velocity information of an aircraft for autonomous navigation. The proposed integrated system is composed of two parts: relative position estimation and absolute position estimation. Relative position estimation recursively computes the current position of an aircraft by accumulating relative displacement estimates extracted from two successive aerial images. Simple accumulation of parameter values reduces the reliability of the extracted parameter estimates as an aircraft goes on navigating, resulting in a large positional error. Therefore, absolute position estimation is required to compensate for the positional error generated by the relative position estimation. Absolute position estimation algorithms using image matching and digital elevation model (DEM) matching are presented. In the image matching, a robust-oriented Hausdorff measure (ROHM) is employed, whereas in the DEM matching, an algorithm using multiple image pairs is used. Experiments with four real aerial image sequences show the effectiveness of the proposed integrated position estimation algorithm     

Color correction for an image sequence

The color correction technique presented constructs a mapping from the registered overlapping regions of two adjacent images in an image sequence. The method involves two main steps. In the first step we use multiple regression to find the relationship between the registered overlapping regions of two adjacent images in the sequence. We then use the derived mapping to correct the color in one image so that it matches the adjacent image. After describing the basic method and discussing strategies far dealing with specific problems that arise, we present and analyze relevant test results     

基于响应面的概率集群优化方法

面对实际应用中的大规模优化问题,基于响应面估计的概率集群优化方法以设计变量的概率分布作为优化对象,而非直接对设计变量值进行优化,可适应连续、离散及混合的设计变量类型。采用响应面构建概率集群评估函数的近似模型,并采用置信区间方法在迭代优化过程中不断更新响应曲面以确保近似精度。实验结果表明算法对解决复杂优化问题有效。     

Dynamic image sequence analysis using fuzzy measures

In this paper, we present an image understanding system using fuzzy sets and fuzzy measures. This system is based on a symbolic object-oriented image interpretation system. We apply a simple, powerful three-dimensional (3-D) recursive filter to tracking moving objects in a dynamic image sequence. This filter has a time-varying 3-D frequency-planar passband that is adapted in a feedback system to automatically track moving objects. However, as objects in the image sequence are not well-defined and are engaged in dynamic activities, their shapes and trajectories in most cases can be described only vaguely. In order to handle these uncertainties, we use fuzzy measures to capture subtle variations and manage the uncertainties involved. This enables us to develop an image understanding system that produces a very natural output. We demonstrate the effectiveness of our system with complex real traffic scenes.     

New results using an integrated model and recursive algorithm for image motion estimation

A linear equation in the affine parameters used to model image motion may be derived by Taylor series expansion and truncation, and windowed spatial integration. Two methods for reducing errors in the Taylor approximation are discussed and results are presented.     

基于SURF特征的航空序列图像位置估计算法

视觉传感器在航空无人机导航和定位任务中应用越来越广泛。针对无人机位置参数估计问题,提出了一种基于SURF特征的图像配准算法,该算法能够适应航空序列图像的旋转、尺度变换及噪声干扰,实现无人机位置的精确估计。构建了SURF尺度空间,运用快速Hessian矩阵定位极值点,计算出航空图像的64维SURF特征描述子;基于Hessian矩阵迹完成特征点匹配;使用RANSAC算法剔除出格点,实现位置参数的精确估计。通过航空图像序列实测数据位置估计实验,验证了该算法的有效性。     

Snow surface albedo estimation using terrestrial photography

A flexible and inexpensive remote sensing tool for albedo estimation using conventional terrestrial photography and its validation on an Alpine glacier are described. The proposed technique involves georeferencing oblique photographs to a digital elevation model (DEM), defining a mapping function between the information contained on a given pixel of the image and the corresponding cell of the DEM. This is attained by performing a perspective projection of the DEM after a viewing transformation into the camera coordinate system. Once the image is georeferenced, the reflectance values recorded by the film or digital camera are corrected for topographic and atmospheric influences and for the effect of the photographic process (lens-film-developing-scanning). Atmospheric transmittance is evaluated using the MODTRAN radiative transfer model. Diffuse and direct irradiation are estimated using a parametric solar irradiation model. The solar-ground geometry, anisotropy of reflected radiation, the effect of surrounding topography and the portion of visible sky are evaluated using terrain algorithms applied to the DEM. The response of the camera-film-scanner system is evaluated using an empirical approach. The result is a geographically correct map of normalized reflectance values. By comparing these to a surface of known albedo, the spatial distribution of albedos is calculated. Comparisons to in situ measurements on the Mer de Glace glacier, French Alps, show good agreement. Sources of error are identified and ways of improvement addressed. The georeferencing algorithm, implemented into the Interactive Data Language (IDL) is available from the author and at the user contributed IDL library at www.rsinc.com.     

A method of model improvement for spotting recognition of gestures using an image sequence

We have developed a real-time gesture recognition system whose models can be taught by only one instruction. Therefore the system can adapt to new gesture performer quickly but it can not raise the recognition rates even if we teach gestures many times. That is because the system could not utilize all the teaching data. In order to cope with the problem, averages of teaching data are calculated. First, the best frame correspondence of the teaching data and the model is obtained by Continuous DP. Next the averages and variations are calculated for each frame of the model. We show the effectiveness of our method in the experiments. Takuichi Nishimura: He is a researcher of Multi-modal Function Tsukuba Laboratory and Information Basis Function Laboratory at the Real World Computing Partnership. He has engaged in motion image understanding, multi-modal human computer interface, multi-modal information retrieval, and mobile robot navigation. He completed the master’s course of the University of Tokyo in 1992. Hiroaki Yabe: He is from SHARP corporation working as a researcher of Multi-modal Function Tsukuba Laboratory and Information Basis Function Tsukuba Laboratory at the Real World Computing Partnership. He has engaged in motion image understanding, multi-modal human computer interface, multi-modal information retrieval. He completed the master’s course of the University of Tokyo in 1995. Ryuichi Oka, Ph.D.: He is a chief of Multi-modal Function Tsukuba Laboratory and Information Basis Function Laboratory at Tsukuba Research Center of the Real World Computing Partnership (RWC Japan) which started in 1992. His research interests include motion image understanding, spontaneous speech understanding, self-organisation information base, multi-modal human computer interface, multi-modal information retrieval, mobile robot, integration of symbol and pattern, and super parallel computation. He received his Ph.D degree in Engineering from the University of Tokyo.     

Face posture estimation using eigen analysis on an IBR (image based rendered) database

In this paper, we present a novel representation of the human face for estimating the orientation of the human head in a two dimensional intensity image. The method combines the use of the much familiar eigenvalue based dissimilarity measure with image based rendering. There are two main components of the algorithm described here: the offline hierarchical image database generation and organization, and the online pose estimation stage. The synthetic images of the subject's face are automatically generated offline, for a large set of pose parameter values, using an affine coordinate based image reprojection technique. The resulting database is formally called as the IBR (or image based rendered) database. This is followed by the hierarchical organization of the database, which is driven by the eigenvalue based dissimilarity measure between any two synthetic image pair. This hierarchically organized database is a detailed, yet structured, representation of the subject's face. During the pose estimation of a subject in an image, the eigenvalue based measure is invoked again to search the synthetic (IBR) image closest to the real image. This approach provides a relatively easy first step to narrow down the search space for complex feature detection and tracking algorithms in potential applications like virtual reality and video-teleconferencing applications.     

Blind image noise level estimation using texture-based eigenvalue analysis

Blind noisy image estimation is useful in many visual processing systems. The challenge lies in accurately estimating the image noise level without any priori information of the image. To tackle this challenge, an iterative texture-based eigenvalue analysis approach is proposed in this paper. The proposed approach utilizes the eigenvalue analysis to mathematically derive a new noise level estimator based on weak-textured image patches. Furthermore, a new texture strength measure is proposed to adaptively select weak-textured patches from the noisy image. Experimental results are provided to demonstrate that the proposed image noise level estimation approach yields superior accuracy and stability performance to that of conventional noise level estimation approaches, so that to improve the performance of image denoising algorithm.     

Depth estimation and image restoration using defocused stereo pairs

We propose a method for estimating depth from images captured with a real aperture camera by fusing defocus and stereo cues. The idea is to use stereo-based constraints in conjunction with defocusing to obtain improved estimates of depth over those of stereo or defocus alone. The depth map as well as the original image of the scene are modeled as Markov random fields with a smoothness prior, and their estimates are obtained by minimizing a suitable energy function using simulated annealing. The main advantage of the proposed method, despite being computationally less efficient than the standard stereo or DFD method, is simultaneous recovery of depth as well as space-variant restoration of the original focused image of the scene.     

Synergistic arc-weight estimation for interactive image segmentation using graphs

We introduce a framework for synergistic arc-weight estimation, where the user draws markers inside each object (including background), arc weights are estimated from image attributes and object information (pixels under the markers), and a visual feedback guides the user’s next action. We demonstrate the method in several graph-based segmentation approaches as a basic step (which should be followed by some proper approach-specific adaptive procedure) and show its advantage over methods that do not exploit object information and over methods that recompute weights during delineation, which make the user to lose control over the segmentation process. We also validate the method using medical data from two imaging modalities (CT and MRI-T1).     

Document image binarization using background estimation and stroke edges

Document images often suffer from different types of degradation that renders the document image binarization a challenging task. This paper presents a document image binarization technique that segments the text from badly degraded document images accurately. The proposed technique is based on the observations that the text documents usually have a document background of the uniform color and texture and the document text within it has a different intensity level compared with the surrounding document background. Given a document image, the proposed technique first estimates a document background surface through an iterative polynomial smoothing procedure. Different types of document degradation are then compensated by using the estimated document background surface. The text stroke edge is further detected from the compensated document image by using L1-norm image gradient. Finally, the document text is segmented by a local threshold that is estimated based on the detected text stroke edges. The proposed technique was submitted to the recent document image binarization contest (DIBCO) held under the framework of ICDAR 2009 and has achieved the top performance among 43 algorithms that are submitted from 35 international research groups.     

Percent surface water estimation from MODIS BRDF 16-day image composites

Four models for deriving percent surface water estimates were developed for use with MODIS16-day Bidirectional Reflectance Distribution Function (BRDF) corrected composite images. The models allow intra-annual surface water estimates to be produced with 1 km spatial resolution and an 8-day temporal resolution when applied to image composites from sensors on both the Aqua and Terra platforms. The surface water models are conceptually simple, relying on widely used indices (NDVI, NDWI, and tasseled cap), but computationally intensive. The models differ in the time and effort required to produce or acquire the inputs necessary for model training. The models were applied and tested in Yukon Flats National Wildlife Refuge, an area with varied surface water types including ponds, fens, and the Yukon River and its tributaries. Resulting accuracies peaked with an R² of approximately 0.625, and model accuracies were higher for pixels with higher percentages of water.     

Range image reconstruction based on robust multiresolution estimation of surface parameters

This paper proposes two robust multiresolution estimation methods of surface parameters for range images. Based on the robust estimation of surface parameters, the proposed methods approximate a patch to a planar surface in the locally adaptive window. Selection of resolution is made pixelwise by comparing a locally computed homogeneity measure with the global threshold obtained by the distribution of the approximation error. The proposed multiresolution surface parameter estimation methods are applied to range image reconstruction and segmentation. Computer simulation results with noisy images contaminated by additive Gaussian noise and impulse noise show that the proposed multiresolution reconstruction methods preserve step and roof edges better than the conventional methods. Also the segmentation methods based on the estimated surface parameters are shown to be robust to noise.     

Non-stationary data sequence classification using online class priors estimation

Online classification is important for real time data sequence classification. Its most challenging problem is that the class priors may vary for non-stationary data sequences. Most of the current online-data-sequence-classification algorithms assume that the class labels of some new-arrived data samples are known and retrain the classifier accordingly. Unfortunately, such assumption is often violated in real applications. But if we were able to estimate the class priors on the test data sequence accurately, we could adjust the classifier without retraining it while preserving a reasonable accuracy. There has been some work on the class priors estimation to classify static data sets using the offline iterative EM algorithm, which has been proved to be quite effective to adjust the classifier. Inspired by the offline iterative EM algorithm for static data sets, in this paper, we propose an online incremental EM algorithm to estimate the class priors along the data sequence. The classifier is adjusted accordingly to keep pace with the varying distribution. The proposed online algorithm is more computationally efficient because it scans the sequence only once. Experimental results show that the proposed algorithm indeed performs better than the conventional offline iterative EM algorithm when the class priors are non-stationary.