The aim of the presented work is to demonstrate enhanced target recognition and improved false alarm rates for a mid to long range detection system, utilising a Long Wave infrared (LWIR) sensor. By exploiting high qua...
详细信息
ISBN:
(数字)9781510603790
ISBN:
(纸本)9781510603783;9781510603790
The aim of the presented work is to demonstrate enhanced target recognition and improved false alarm rates for a mid to long range detection system, utilising a Long Wave infrared (LWIR) sensor. By exploiting high quality thermal image data and recent techniques in machine learning, the system can provide automatic target recognition capabilities. A Convolutional Neural Network (CNN) is trained and the classifier achieves an overall accuracy of > 95% for 6 object classes related to land defence. While the highly accurate CNN struggles to recognise long range target classes, due to low signal quality, robust target discrimination is achieved for challenging candidates. The overall performance of the methodology presented is assessed using human ground truth information, generating classifier evaluation metrics for thermal image sequences.
A dim moving targetdetection algorithm based on spatio-temporal classification sparse representation, which can characterize the motion information and morphological feature of target and background clutter, is propo...
详细信息
A dim moving targetdetection algorithm based on spatio-temporal classification sparse representation, which can characterize the motion information and morphological feature of target and background clutter, is proposed to enhance the performance of targetdetection. A spatio-temporal redundant dictionary is trained according to the content of infrared image sequence, and then is subdivided into target spatio-temporal redundant dictionary describing moving target, and background spatio-temporal redundant dictionary embedding background by the criterion that the target spatio-temporal atom could be decomposed more sparsely over Gaussian spatio-temporal redundant dictionary. The target and background clutter can be sparsely decomposed over their corresponding spatio-temporal redundant dictionary, yet could not be sparsely decomposed on their opposite spatio-temporal redundant dictionary, and so their residuals after reconstruction by the prescribed number of target and background spatio-temporal atoms would differ very visibly. Some experimental results show this proposed approach could not only improve the sparsity more efficiently, but also enhance the targetdetection performance more effectively. (C) 2014 Elsevier B.V. All rights reserved.
Optoelectronic imaging system which loaded on ships have several imaging channels generally. They can cover visible light (daylight and low-light), medium wave infrared and long wave infrared. To that systems have onl...
详细信息
ISBN:
(纸本)9781628419009
Optoelectronic imaging system which loaded on ships have several imaging channels generally. They can cover visible light (daylight and low-light), medium wave infrared and long wave infrared. To that systems have only two channels, visible light imaging is kept. In this paper, for the requirement of target real-time detection and classification under sea-sky background, image data from different channels are processed independently using Harris feature of targets and texture feature of background, then the result data from different channels are fused and compared to delete fake targets and interference from background, in order to reduce false alarm rate and improve the detection location precision. Based on the location relationship between targets and different type background (sky or sea), the target types are determined. For the rigidly requirement of system real time, multithread mechanism and big neighborhood processor are applied for parallel data processing, in order to reduce the processing time less than one frame time. At last, the experiment has been done on two channels system, approving that the method in this paper can improve the comprehensive searching performance of optoelectronic imaging system.
The confidence of targetdetection can be used to evaluate the reliability and risk level of the detected targets and can effective help to exclude the false alarms, but very little investigation was involved in the p...
详细信息
The confidence of targetdetection can be used to evaluate the reliability and risk level of the detected targets and can effective help to exclude the false alarms, but very little investigation was involved in the past. In this letter, a confidence-driven infraredtargetdetection method is proposed. We develop three confidence evaluating methods: (1) the median classification confidence of the cascade classifier;(2) the context confidence based on the number and the confidence of the merged detection rectangles around the detected target;and (3) the contrast confidence based on the difference between the detected target distribution and the around background distribution. The three confidences are combined to form the final confidence of the detected targets. We then use the confidence to refine the localization of the targets. The evaluation using real infrared images demonstrates the good performance of the proposed confidence-driven infrareddetection algorithm on both undetected error and false alarm. (C) 2014 Elsevier B.V. All rights reserved.
target tracking within conventional video imagery poses a significant challenge that is increasingly being addressed via complex algorithmic solutions. The complexity of this problem can be fundamentally attributed to...
详细信息
ISBN:
(纸本)9781628413168
target tracking within conventional video imagery poses a significant challenge that is increasingly being addressed via complex algorithmic solutions. The complexity of this problem can be fundamentally attributed to the ambiguity associated with actual 3D scene position of a given tracked object in relation to its observed position in 2D image space. We propose an approach that challenges the current trend in complex tracking solutions by addressing this fundamental ambiguity head-on. In contrast to prior work in the field, we leverage the key advantages of thermal-band infrared (IR) imagery for the pedestrian localization to show that robust localization and foreground target separation, afforded via such imagery, facilities accurate 3D position estimation to within the error bounds of conventional Global Position System (GPS) positioning. This work investigates the accuracy of classical photogrammetry, within the context of current targetdetection and classification techniques, as a means of recovering the true 3D position of pedestrian targets within the scene. Based on photogrammetric estimation of target position, we then illustrate the efficiency of regular Kalman filter based tracking operating on actual 3D pedestrian scene trajectories. We present both a statistical and experimental analysis of the associated errors of this approach in addition to real-time 3D pedestrian tracking using monocular infrared (IR) imagery from a thermal-band camera.
This paper presents a methodology and results for the comparison of simulated imagery to real imagery acquired with multiple sensors hosted on an airborne platform. The dataset includes aerial multi- and hyperspectral...
详细信息
ISBN:
(纸本)9781628410259
This paper presents a methodology and results for the comparison of simulated imagery to real imagery acquired with multiple sensors hosted on an airborne platform. The dataset includes aerial multi- and hyperspectral imagery with spatial resolutions of one meter or less. The multispectral imagery includes data from an airborne sensor with three-band visible color and calibrated radiance imagery in the long-, mid-, and short-wave infrared. The airborne hyperspectral imagery includes 360 bands of calibrated radiance and reflectance data spanning 400 to 2450 nm in wavelength. Collected in September 2012, the imagery is of a park in Avon, NY, and includes a dirt track and areas of grass, gravel, forest, and agricultural fields. A number of artificial targets were deployed in the scene prior to collection for purposes of targetdetection, subpixel detection, spectral unmixing, and 3D object recognition. A synthetic reconstruction of the collection site was created in DIRSIG, an image generation and modeling tool developed by the Rochester Institute of technology, based on ground-measured reflectance data, ground photography, and previous airborne imagery. Simulated airborne images were generated using the scene model, time of observation, estimates of the atmospheric conditions, and approximations of the sensor characteristics. The paper provides a comparison between the empirical and simulated images, including a comparison of achieved performance for classification, detection and unmixing applications. It was found that several differences exist due to the way the image is generated, including finite sampling and incomplete knowledge of the scene, atmospheric conditions and sensor characteristics. The lessons learned from this effort can be used in constructing future simulated scenes and further comparisons between real and simulated imagery.
暂无评论