• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Conference on Targets and Backgrounds IX: Characterization and Representation Apr 21-22, 2003 Orlando, Florida, USA >COMPARISON OF HUMAN AND ALGORITHMIC TARGET DETECTION IN PASSIVE INFRARED IMAGERY
【24h】

COMPARISON OF HUMAN AND ALGORITHMIC TARGET DETECTION IN PASSIVE INFRARED IMAGERY

机译：被动红外影像中人类和算法目标检测的比较

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

We have designed an experiment that compares the performance of human observers and a scale-insensitive target detection algorithm for the detection of ground targets in passive infrared imagery. Though the algorithm performed at considerably higher false alarm rates as compared to human observers, detection was comparable, and it did so using simple pixel level features only. The test database contains targets near clutter whose detectability ranged from easy to very difficult. Results indicate that human observers detect more "easy-to-detect" targets, and with far fewer false alarms, than the algorithm. For "difficult-to-detect" targets, human and algorithm detection rates are considerably degraded, and algorithm false alarms excessive. Analysis of detections as a function of observer confidence shows that algorithm confidence attribution does not correspond to human attribution, and does not adequately correlate with correct detections. The best target detection score for any human observer was 84%, as compared to 55% for the algorithm for the same false alarm rate. At 81%, the maximum detection score for the algorithm, the same human observer had 6 false alarms per frame as compared to 29 for the algorithm. Individually, human observers detected no more than 84% of the targets; whereas, taken collectively the seven observers tested detected 96% of the targets. Analysis of performance as a function of range to target indicates that for both algorithm and human observers, performance degrades with increasing range to target, and that at any level of detection and range human false alarm rejection is at least 10 times better than that of the algorithm. Although no human, infrared imagery, experts were used for this testing the performance of the seven human observers shows that human observers with little to no previous experience performed about as well as observers with extensive previous experience. Detector ROC curves and observer-confidence analysis benchmarks the algorithm and provides insights into algorithm deficiencies and possible paths to improvement.
机译：我们设计了一个实验，该实验比较了人类观察者的性能和一种对比例尺不敏感的目标检测算法，用于检测被动红外图像中的地面目标。尽管与人类观察者相比，该算法的误报率要高得多，但检测是可比的，并且仅使用简单的像素级功能就可以做到。测试数据库包含杂乱无章的目标，可检测性范围从容易到非常困难。结果表明，与该算法相比，人类观察者可以检测到更多“易于检测”的目标，并且虚假警报要少得多。对于“难以检测”的目标，人和算法的检测率大大降低，并且算法错误警报过多。对作为观察者置信度函数的检测分析表明，算法置信度归因与人的归因不符，并且与正确的检测不充分相关。对于任何人类观察者，最佳目标检测得分为84％，而对于相同的误报率，该算法为55％。该算法的最高检测分数为81％，同一人类观察者每帧出现6次虚警，而算法为29次。个别而言，人类观察者发现的目标不超过84％；相反，接受测试的七名观察员则总共发现了96％的目标。对作为目标范围的函数的性能分析表明，对于算法观察者和人类观察者而言，性能都会随着目标范围的增大而降低，并且在任何检测水平和范围内，人为错误警报拒绝的性能至少要比目标范围高10倍。算法。尽管没有人的红外图像，但专家们进行了这项测试，这七个人的观察员的表现表明，以前观察很少或没有经验的人观察员的表现与以前有丰富经验的观察员的表现差不多。检测器ROC曲线和观察者信心分析对算法进行了基准测试，并提供了有关算法缺陷和改进途径的见解。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号