摘要
建立适合煤矿井下特殊环境的危险区域目标检测系统结构和新的目标匹配算法。新算法基于SIFT(scale-invariant feature transform)多尺度变换,结合形态学技术用降维后的局部区域匹配方法提高系统实时性;交叉匹配粗筛选后将RANSAC(random sample consensus)算法和L-M(Lev-enberg-Marquardt)非线性优化算法结合估计优化参数,解决现有算法计算复杂,匹配时间长,复杂环境匹配精度低的问题。实验证明,新算法对煤矿井下模糊、低照度、遮挡、高噪声和尺度变化等情况均具有良好的鲁棒性,解决多摄像机不同视角目标匹配问题,适合实时处理的监控系统中井下危险区域目标检测。
Established underground dangerous areas object detection system architecture and new object detecting algorithm applied on coal mine special areas.New algorithm based on SIFT（scale-invariant feature transform）,used region matching and dimension reduction methods combined with morphology pretreatment technique to advance system real time ability.After cross-matching cursorily,new algorithm combined RANSAC（random sample consensus） and L-M（Levenberg-Marquardt） nonlinear optimization algorithms to estimate optimization parameters,solved the now algorithms＇ high computational complexity,long matching time and low matching accuracy problems in complex environment.The results show that the new algorithm has good robustness on blur,low illumination,shelter by other object,high noise and scale change condition.It solves the matching problem under multi-camera with different visual angle,suits for real-time processing of video surveillance and object detecting in dangerous areas of coal mine.
出处
《煤炭学报》
EI
CAS
CSCD
北大核心
2011年第3期527-532,共6页
Journal of China Coal Society
基金
国家高技术研究发展计划(863)重点资助项目(2008AA062200)
江苏省产学研联合创新资金项目(BY2009114)
