Differential inter-layer adaptive fusion change detection network

The paper presents an automatic and efficient video object segmentation algorithm. Moving object extraction is carried out by adaptive background model and edge-based change detection techniques. The background is updated by use of pixel history and a moving object mask. Connected component analysis and morphological filtering are employed to obtain an accurate VOP (video object plane). Finally, a novel object tracking window scheme is applied to improve the processing speed. Experimental results for three different types of MPEG-4 video sequences are shown to demonstrate the effectiveness of the proposed algorithm

Shot change detection is a main technique for automatic temporal segmentation of video, which requires adaptive adjustment of thresholds and real-time operation in actual applications. To date, few researches have been conducted on real-time shot change detection to be applied to hardware terminals with low performance such as PMPs (portable media player) and cellular phones. In this paper, we propose an adaptive real-time shot change detection algorithm and implement it on an actual PMP. The proposed algorithm consists of the following elements: sub-sampling, weighting variance, and adaptive thresholds. Our experiments obtained the detection rate of about 94.4% in overall accuracy, demonstrating a higher detection rate and a computational amount 1/256 less than the conventional algorithms which have optimal fixed thresholds adjusted by humans for each video. We verified the real-time operation of shot change detection by implementing our algorithm on the PMP of Homecast, a PMP company. The proposed algorithm will be helpful in searching video data on portable media players such as PMPs and cellular phones.

An adaptive model for online detection of relevant state changes in Internet-based systems

This paper is concerned with change detection in averaged multi-look SAR imagery. Averaged multi-look SAR images are preferable to full aperture SAR reconstructions when the imaging algorithm is approximation based (e.g., polar format processing), or motion data are not accurate over a long full aperture. We study the application of a SAR change detection method, known as signal subspace processing, that is based on the principles of 2D adaptive filtering (M. Soumekh, January 1999) and we use it to recognize the addition of surface landmines to a particular area under surveillance. We identify the change detection problem as a trinary hypotheses testing problem, and identify a change signal and its normalized version to determine whether there is i) no change in the imaged scene; ii) a target has entered the imaged scene; or iii) a target has exited the imaged scene. A statistical analysis of the error signal is provided to show its properties and merits. Results are provided with a realistic X band SAR platform using averaged noncoherent multi-look and coherent single-look SAR imagery.

Hyperspectral change detection, which provides abundant information on land cover changes in the Earth's surface, has become one of the most crucial tasks in remote sensing. Recently, deep-learning-based change detection methods have shown remarkable performance, but the acquirement of labeled data is extremely expensive and time-consuming. It is intuitive to learn changes from the scene with sufficient labeled data and adapting them into an unlabeled new scene. However, the nonnegligible domain shift between different scenes leads to inevitable performance degradation. In this article, a cycle-refined multidecision joint alignment network (CMJAN) is proposed for unsupervised domain adaptive hyperspectral change detection, which realizes progressive alignment of the data distributions between the source and target domains with cycle-refined high-confidence labeled samples. There are two key characteristics: 1) progressively mitigate the distribution discrepancy to learn domain-invariant difference feature representation and 2) update the high-confidence training samples of the target domain in a cycle manner. The benefit is that the domain shift between the source and target domains is progressively alleviated to promote change detection performance on the target domain in an unsupervised manner. Experimental results on different datasets demonstrate that the proposed method can achieve better performance than the state-of-the-art change detection methods.

Nowadays, the advance of the technology allows robots to acquire dense point clouds decreasing the price and increasing the performance. However, it is a hard task to deal with due to the large amount of points, the redundancy and the noise. This paper proposes an adaptable system to build a 3D feature model of point clouds using Gaussian Mixture Models. These 3D models are used in order to detect changes in the autonomous robot's working environment. The presented work describes an efficient change detection system based on two consecutive stages. First, a top-down approach estimates features using Gaussian Mixture Models. The presented new approach improves the performance of previous related works in terms of computational load and robustness, nevertheless the system is selection criteria dependent. Thus, the efficiency of different selection criteria are evaluated and compared in this paper. Experimental results demonstrate that the Minimum Distance Length (MDL) criteria outperforms the other studied methods. In the second stage, a change detection method is performed using the previously estimate Mixture of Gaussians. The proposed full system is able to detect changes using Gaussian Mixture Models with a reduced computational cost in relation to state-of-art algorithms.

In this paper, we propose and illustrate a methodology for classifying the change detection results generated from repeatpass polarimetric RADARSAT-2 images and segmenting only the changes of interest to a given user while suppressing all other changes. The detected changes are first classified based on generated supervised ground-cover classification of the polarimetric SAR images between which changes were detected. In the absence of reliable ground truth needed for generating supervised classification training sets, we rely on the use of periodically acquired high-resolution, multispectral optical imagery in order to classify the manually selected training sets before computing their classes' statistics from the SAR images. The classified detected changes can then be segmented to isolate the changes of interest, as specified by the user and suppress all other changes. The proposed polarimetric change detection, classification and segmentation method overcomes some of the challenges encountered when visualizing and interpreting typical raw change results. Often these non-classified change detection results tend to be too crowded, as they show all the changes including those of interest to the user as well as other non-relevant changes. Also, some of the changes are difficult to interpret, especially those which are attributed to a mixture of the backscatters. We shall illustrate how to generate, classify and segment polarimetric change detection results from two SAR images over a selected region of interest.

A novel technique for anomalous change detection in hyperspectral images is presented. It adaptively measures the spectral distance between corresponding pixels in multi-temporal images by exploiting the local estimates of the signal to noise ratio for each spectral component of the pixel under test. Different metrics have been compared, based on multidimensional angular distance. Results obtained by applying the new algorithm to real data are presented and discussed. Performance evaluation highlighted the effectiveness of the proposed approach with respect to traditional methods, resulting in a consistent improvement of both the probability of detection of changes and the capability of suppressing the background.

Two adaptive algorithms are presented for the detection of small changes in a pair of images in a low signal to clutter plus noise ratio (SCNR) environment. They both have the ability to track the nonstationary image signals and suppress the clutter plus noise background. Both detectors are based on the adaptive correlation canceling technique. One algorithm uses an order recursive least squares (ORLS) lattice filter, while the other is based on the two-dimensional least mean square (TDLMS) algorithm. The only a priori information required by the algorithms is that the background clutter plus noise in the pair of images is spatially correlated. An analytical expression for the improvement factor for the change detectors is presented. The performance of the two algorithms is evaluated by using an optical satellite image, with computer generated target and noise added. >

When we consider the advance in spatial resolution of remote sensing images, there is a potential demand for spatial segmentation based on spatial pattern dynamics. In this paper, we propose a spatial pattern segmentation method as a spatial version of adaptive filtering and change detection of temporal process. Our prosal detects boundaries between adjacent spatial clusters by evaluating the predictability of adaptive filters. Pre- liminary experiments show significant validity for the boundary detection. In addition, we propose a critical change detection method as temporal version of our spatial segmentation method which evaluates predictablility based on transition probability matrix. An effective critical change detection is realized by the combination of the evaluation of (1) the transition probability from a normal state to singular state; and (2) the stability of singular state.

An adaptive filter is derived in a Bayesian framework from the assumption that the difference in the parameter distribution from one time to another is bounded in terms of the Kullback-Leibler divergence. We show an explicit link to the general concepts of exponential forgetting, and outline the details for a linear Gaussian model with unknown parameter and covariance. We extend the problem to an unknown forgetting factor, where we provide a particular prior that allows for abrupt changes in forgetting, which is useful in change detection problems. The Rao-Blackwellized particle filter is used for the implementation, and its performance is assessed in a simulation of system with abrupt changes of parameters.

There has been a keen interest in detecting abrupt sequential changes in streaming data obtained from sensors in wireless sensor networks for Internet of Things applications, such as fire/fault detection, activity recognition, and environmental monitoring. Such applications require (near) online detection of instantaneous changes. This paper proposes an online, adaptive filtering-based change detection (OFCD) algorithm. Our method is based on a convex combination of two decoupled least mean square windowed filters with differing sizes. Both filters are applied independently on data streams obtained from sensor nodes such that their convex combination parameter is employed as an indicator of abrupt changes in mean values. An extension of our method (OFCD) based on a cooperative scheme between multiple sensors (COFCD) is also presented. It provides an enhancement of both convergence and steady-state accuracy of the convex weight parameter. Our conducted experiments show that our approach can be applied in distributed networks in an online fashion. It also provides better performance and less complexity compared with the state-of-the-art on both of single and multiple sensors.

Detection of structural changes from an operational process is a major goal in machine condition monitoring. Existing methods for this purpose are mainly based on retrospective analysis, resulting in a large detection delay that limits their usages in real applications. This paper presents a new adaptive real-time change detection algorithm, an extension of the recent research by combining with an incremental sliding-window strategy, to handle the multi-change detection in long-term monitoring of machine operations. In particular, in the framework, Hilbert space embedding of distribution is used to map the original data into the Re-producing Kernel Hilbert Space (RKHS) for change detection; then, a new adaptive threshold strategy can be developed when making change decision, in which a global factor (used to control the coarse-to-fine level of detection) is introduced to replace the fixed value of threshold. Through experiments on a range of real testing data which was collected from an experimental rotating machinery system, the excellent detection performances of the algorithm for engineering applications were demonstrated. Compared with state-of-the-art methods, the proposed algorithm can be more suitable for long-term machinery condition monitoring without any manual re-calibration, thus is promising in modern industries.

The proposed system is aimed at detecting and classifying 3D moving objects for security control of unmanned automatic railway stations. Most common approaches are based on active sensors like optical barriers or laser scanning devices. The proposed approach, named 3DMODS, is based on stereo vision technology, using a prediction-verification paradigm. Adaptive change detection is performed at the video rate to detect immediately moving objects in the scene. Object features are collected by "scanning" the scene with different parallel planes at variable height, to verify the volumetric consistency of the detected object. A prediction of stereo correspondence is performed, using homographic transformation on the set of predefined 3D planes, to verify whether the detected change is really a moving 3D object with a significant size, or just a phantom caused by shadows or highlights. A simple classification scheme is currently implemented to decide for an alarm candidate in case of relevant object size, but much more complex and flexible solutions are possible, to recognize all the relevant objects in the scene and achieve much more robust alarm detection performance.

High-gain observer-based parameter identification with application in a gas turbine engine