Showing papers on "Change detection published in 1992"
01 Jan 1992
TL;DR: This paper shows here that this approach based on the so-called “asymptotic local” approach for change detection is of much wider applicability: model reduction can be enforced, biased identification procedures can be used, and finally one can even get rid of identification and use instead some much simpler Monte-Carlo estimation technique prior to change detection.
Abstract: Techniques for early warning of slight changes in systems and plants are useful for condition based maintenance. In this paper we present an approach for this problem. This approach is based on the so-called "asymptotic local" approach for change detection previously introduced by the same authors. Its original principle consists in characterizing a system via some identified model, and then to monitor its changes using some data-to-model distance also derived from identification technique. We show here that this method is of much wider applicability : model reduction can be enforced, biased identification procedures can be used and finally one can even get rid of identification and use instead some much simple Monte-Carlo estimation technique prior to change detection. Experiments on AR model are reported and an example from gas turbine industry is briefly discussed.
49 citations
TL;DR: In this article, a change detection technique was applied to map urban expansion in Hong Kong from 1979 to 1987 with a Landsat MSS and a SPOT HRV data, and a thresholding technique was employed to separate the areas of changes from those of no-change.
Abstract: The objective of this paper is to map urban expansion in Hong Kong from 1979 to 1987 with a Landsat MSS and a SPOT HRV data. The data were radiometrically calibrated and geometrically registered. Three change detection techniques were applied. First, image overlay was used to enhance change areas visually. Second, a standardized principal components analysis was performed to yield minor components which were change related vectors. A thresholding technique was employed to separate the areas of changes from those of no-change. A binary change mask was created. Third, a post-classification comparison was merged with the change mask to identify the nature of specific land use and land cover changes. Major land development in the city can be easily detected and mapped with these techniques.
25 citations
TL;DR: Some results of recent work on development and investigation of the sequential change detection algorithm based upon a likelihood ratio are presented.
19 citations
26 May 1992
TL;DR: In this paper, the results of the calibration analysis performed to date on ASF-produced ERS-1 SAR images, and some preliminary results on change detection on the Alaskan North Slope derived from the same.
Abstract: The following paper describes, briefly, the results of the calibration analysis performed to date on ASF-produced ERS-1 SAR images, and and some preliminary results on change detection on the Alaskan North Slope derived from the same. Image quality, geometric and radiometric fidelity and repeat pass radiometric stability have all been determined to be satisfactory. A calibration workstation has been designed and implemented for use in operational data quality analysis of ASF data products. In addition, higher level data analysis programs have been developed and are being used for interferometry, texture analysis, and change detection.
17 citations
26 May 1992
TL;DR: The structure and initial results of a symbolic, model-based approach to feature classification in Synthetic Aperture Radar (SAR) imagery is revealed, with initial testing in a 71% correct classification rate in a fully automated mode.
Abstract: The structure and initial results of a symbolic, model-based approach to feature classification in Synthetic Aperture Radar (SAR) imagery is pre- sented. The. models include quantitative, qualitative and relational as- pects of feature signatures. The prototype system consists of a combina- tion of a 120 rule knowledge-based analysis tools, and a 20 procedure data extraction library. The initial testing over the four SAR data set re- sulted in a 71% correct classification rate in a fully automated mode.
12 citations
26 May 1992
10 citations
Patent•
[...]
08 Dec 1992
TL;DR: In this paper, a scene change detection section detects the scene change from the difference between a preceding and a current picture frame, and a step size decision section 18 revises step sizes at each of quantization circuits 5, 7 by a prescribed amount and applies coding to the result.
Abstract: PURPOSE:To attain excellent picture transmission even in a scene change by detecting and processing the scene change based on a difference between a preceding and a current picture frame. CONSTITUTION:An input picture data is stored in a frame memory 19 and outputted to a subtractor 20 when a succeeding frame is inputted. A difference between a preceding and a current picture is obtained and a scene change detection section 21 detects the production of the scene change from the difference based on a prescribed processing condition. When any scene change is detected, a step size decision section 18 revises step sizes Q1, Q2 at each of quantization circuits 5, 7 by a prescribed amount and applies coding to the result. Thus, it is improved that a defect is caused due to an excess code quantity at the scene change.
8 citations
23 Mar 1992
TL;DR: A method is proposed for detecting the quantifying eventual changes in radar backscatter and mapping out ensembles of pixels of spatially and radiometrically homogeneous and similar changes using a Bayes classifier.
Abstract: Multitemporal synthetic aperture radar (SAR) observations of natural areas permit monitoring of the characteristics and evolution of structural and electrical properties of natural surfaces as a result of meteorological and phenologic cycles. Change detection is one possible way of analyzing multitemporal SAR data which aims at quantifying the relative changes in radar backscatter of the same surface recorded at two different times under the same imaging conditions. A method is proposed for detecting the quantifying eventual changes in radar backscatter and mapping out ensembles of pixels of spatially and radiometrically homogeneous and similar changes using a Bayes classifier. Examples using real multitemporal SAR data are given. >
7 citations
6 citations
Journal Article•
5 citations
Proceedings Article•
01 Jan 1992TL;DR: The results of a calibration analysis performed on ERS-1 synthetic aperture radar (SAR) images produced by the Alaska SAR facility (ASF) are presented, together with some preliminary results on change detection on the Alaskan north slope derived from the same images as mentioned in this paper.
Abstract: The results of a calibration analysis performed on ERS-1 synthetic aperture radar (SAR) images produced by the Alaska SAR facility (ASF) are presented, together with some preliminary results on change detection on the Alaskan north slope derived from the same images. Image quality, geometric and radiometric fidelity, and repeat pass radiometric stability have all been determined to be satisfactory. A calibration workstation has been designed and implemented for use in operational data quality analysis of ASF data products. Higher-level data analysis programs have been developed for interferometry, texture analysis, and change detection.
01 Jan 1992
TL;DR: The VISDTA system as discussed by the authors is an automatic scanning, wide panhilt platform, a processing and control subsystem, and area, surveillance sensor with built-in change detection and an operator console.
Abstract: VISDTA Svstem DescriDtion The Video Imaging System for Detection, Tracking, The VISDTA hardware consists of the sensors and and Assessment (VISDTA) is an automatic scanning, wide panhilt platform, a processing and control subsystem, and area, surveillance sensor with built-in change detection and an operator console. Figure 1 shows the prototype video motion detection features. Several of these thermal demonstration system. Previous reports have presented imager-based systems have been deployed worldwide for demonstration, evaluation, and improvement of site security. Recent modifications and performance improvements are described in terms of environmental and mechanical ruggedness, and detection probability. In addition, a VISDTA expansion is described that could convert many fixed-view cameras into change-detection sensors and further into video motion detection (VMD) devices. This modification proposes easy addition of VMD capability to many existing installations. A summary is included of the recent efforts to transfer VISDTA technology to industry. detailed descriptions of the system hardware [2, 31.
14 Oct 1992
TL;DR: A VISDTA expansion is described that could convert many fixed-view cameras into change-detection sensors and further into video motion detection (VMD) devices, and proposes easy addition of VMD capability to many existing installations.
Abstract: The video imaging system for detection, tracking, and assessment (VISDTA) is an automatic scanning, wide area, surveillance sensor with built-in change detection and video motion detection features. Recent modifications and performance improvements are described in terms of environmental and mechanical ruggedness, and detection probability. In addition, a VISDTA expansion is described that could convert many fixed-view cameras into change-detection sensors and further into video motion detection (VMD) devices. This modification proposes easy addition of VMD capability to many existing installations. A summary is included of the efforts to transfer VISDTA technology to industry. >
01 Feb 1992
TL;DR: The initial testing of the 120 rule prototype over the SAR data sets resulted in a 71% correct classification rate overall and the intended Phase HI classification system will consist of an automated model-driven linked to an image understanding library such that during mated classification, the model-based processor will call various library routines as required to reach solutions.
Abstract: : The feasibility a model-based approach to feature classification and change detection in synthetic aperture radar imagery is investigated. The models include quantitative, qualitative and relational aspects of feature signatures. The intended Phase HI classification system will consist of an automated model-driven linked to an image understanding library such that during mated classification, the model-based processor will call various library routines as required to reach solutions. Phase I effort consisted of: (1) Investigation of various methodologies for rule-oriented model-based approaches; (2) Development and test results of a numerical scoring function for tracking uncertainties during rule-based processing; (3) Developmental and specification of the required image processing library functions; (4) Development and testing of a 120 rule prototype classification system; and (5) Results from testing the combination rule prototype and image processing library on four SAR images. The initial testing of the 120 rule prototype over the SAR data sets resulted in a 71% correct classification rate overall.
TL;DR: In this article, the authors present a method for solving the change point detection problem for ARMA systems which are assumed to have a slow and non-decaying drift after the change occurs.
01 Jan 1992
TL;DR: Two design prototypes of neural network for on-line process chnage detection in dynamic model-switching environments are presented and two important design considerations are discussed; i.e., what type of distance measures are suitable for process change detection and how the neural networks should be trained foron-line applications.
Abstract: Two design prototypes of neural network for on-line process chnage detection in dynamic model-switching environments are presented The process model is assumed to follow either an ARMAX or a non-linear Volterra representation The connection strengths of the first neural network are adaptable parameter estimates of the underlying process model The second prototype is based on the learning vector quantization procedure Two important design considerations are discussed; ie, what type of distance measures are suitable for process change detection and how the neural networks should be trained for on-line applications Various change detection measures and training procedures are also discussed Finally, on-line performance of the proposed neural networks is demonstrated via computer simulation experiments
Patent•
08 Dec 1992
TL;DR: In this article, the run length detection circuit detects a run color change point in one word for each block, the barrel shifter 3 shifts the same run color part for a succeeding clock and then the succeeding run colour change point and run length between the change points is digitized and the results are accumulated.
Abstract: PURPOSE:To improve the detection speed of a run length by digitizing the run length between change points of a run color in one word and accumulating them. CONSTITUTION:A picture element data of one word is latched to a data buffer device 1 and inputted to a change point detection circuit 7 via a barrel shifter 3. A change point is detected based on a run color signal RC from a control circuit 2 and signals A0-A7, are outputted to an encoder 11 only at the change point. Then, the change point is digitized by the encoder 11 and the result is inputted to a run length accumulator 15 and a shift adder 5. Thus, the run length detection circuit detects a run color change point in one word for each block, the barrel shifter 3 shifts the same run color part for a succeeding clock and then the succeeding run color change point and the run length between the change points is digitized and the results are accumulated thereby improving the run length detection speed.
Proceedings Article•
12 Oct 1992
TL;DR: In this article, the same scene viewed at a series of grazing angles ranging from 1 degrees to 10 degrees was analyzed and the effects of small grazing angle changes to be evaluated, and the implications of these for the robustness of applications such as change detection were discussed.
Abstract: A problem is associated with the repeatability of the radar geometry at which the two scenes are imaged. If there is a significant difference between the two geometries, then the scattering properties between the two views will be different resulting in an increased probability of false detection (of changes). This paper examines the extent to which this occurs by analysing the same scene viewed at a series of grazing angles ranging from 1 degrees to 10 degrees . The authors briefly describe the processing techniques used to provide high quality imagery and the change detection technique itself. They describe the experimentation performed to enable the effects of small grazing angle changes to be evaluated, and present the results of experimental observations. The implications of these for the robustness of applications such as change detection are discussed.
04 Oct 1992
TL;DR: In this article, a systematic approach to select and optimize kernel functions of a time-frequency transformation for parameter change detection in linear, piecewise-constant-parameter systems is presented.
Abstract: A systematic approach to selecting and optimizing kernel functions of a time-frequency transformation for parameter change detection in linear, piecewise-constant-parameter systems is presented. The time-frequency transformation is applied to the system output, and the local moments' sensitivity with respect to the change is maximized. The local moments' optimization leads to a partial characterization of the transformation kernel function. Under proper kernel constraints, the local moments are simply related to the characteristic parameters of a linear system (e.g., natural frequencies, time constants, damping coefficients) and thus are suitable for parameter change diagnosis. An illustrative numerical example based on a second-order linear system is given. >
TL;DR: Tomography with minimum feature constraints provides improved resolution for problems involving detection of anomalous zones in a known background and potentially appropriate applications include tunnel, ore-body, or localized lithology change detection.
Abstract: A fundamental premise of geologic interpretation is that no feature should be present on an interpretation which is not required to explain available data. This may be interpreted as a minimum feature constraint. Tomography with minimum feature constraints provides improved resolution for problems involving detection of anomalous zones in a known background. Potentially appropriate applications include tunnel, ore-body, or localized lithology change detection.
Patent•
03 Apr 1992
TL;DR: In this paper, a serial monitor frame data with a prescribed state decision data is used to detect state change in a CPU by detecting the presence or absence of the state change according to coincidence or non-coincidence.
Abstract: PURPOSE:To shorten time required for detecting a state change in a CPU by comparing a serial monitor frame data with a prescribed state decision data, detecting the presence / absence of the state change according to coincidence / noncoincidence and fetching only the monitor frame data having the state change into the CPU CONSTITUTION:According to a receiving clock, a counter 1 generates the read address signal of a memory 2, and the state decision data decided in advance to be a base is successively outputted from the memory 2 to a comparison part 3 In the comparison part 3, the state decision data is compared with the received serial monitor frame data and in the case of noncoincidence, a state change detection signal is outputted to show the presence of the state change In a CPU 4 receiving this state change detection signal, it is possible by fetching the monitor frame data and processing it to analyze the condition of the state change on the side of transmitting the data or to report the contents of the state change to another device and so on Thus, time for detecting the state change in the CPU 4 is shortened
01 Jan 1992
TL;DR: In this article, a systematic approach to select and optimize kernel function of a time-frequency transformation for parameter change detection in linear, piecewise- constant-parameter systems is presented, where the local moments are simply related to the characteristic parameters of a linear system (e.g. natural frequencies, time-constants, damping coefficients etc.).
Abstract: This paper presents a systematic approach to selecting and optimizing kernel function of a time-frequency transformation for parameter change detection in linear, piecewise- constant-parameter systems. The time-frequency transformation is applied to the system output and the local moments sensitivity with respect to the change is maximized. The local moments optimization leads to a partial characterization of the transformation kernel function. Under proper kernel constraints the local moments are simply related to the characteristic parameters of a linear system (e.g. natural frequencies, time-constants, damping coefficients etc.) and thus are suitable for parameter change diagnosis. An illustrative numerical example based on a second order linear system follows.