基于改进CV模型的煤矿井下早期火灾图像分割
发布时间：2018-09-10 18:17
【摘要】：煤矿井下早期火灾图像中火焰区域、火焰余辉及非火焰高灰度干扰区域三者的灰度值十分接近,利用传统的Chan-Vese(CV)模型很难将火焰区域精确地提取出来。针对这一问题,提出了一种改进的CV模型以实现煤矿井下早期火灾图像的精确分割。在计算目标和背景区域拟合中心时,引入自适应权值进行加权平均,充分考虑了像素点灰度值与拟合中心的差异,并据此确定该点对拟合中心的贡献度,更加精确地计算目标和背景区域的拟合中心;为了加速模型的演化,引入曲线内外区域像素的中值绝对差,替换模型中的内外区域能量系数,提高模型分割效率。最终达到快速提取早期火灾图像中火焰区域的目的。大量实验结果表明,与现有的Otsu算法、CV模型、引入能量权重的CV模型、引入梯度信息的CV模型以及两种类似提出模型的CV模型相比,利用改进CV模型对煤矿井下早期火灾图像,能取得更好的分割效果,并且满足实时性要求。
[Abstract]:The grayscale values of flame region, flame afterglow and non-flame high gray level interference region in early fire images of underground coal mine are very close, so it is difficult to extract the flame region accurately by using traditional Chan-Vese (CV) model. To solve this problem, an improved CV model is proposed to achieve accurate segmentation of early fire images in coal mines. When calculating the fitting center of the target and background region, the adaptive weight value is introduced to weighted average, and the difference between the pixel gray value and the fitting center is fully considered, and the contribution of the point to the fitting center is determined according to the difference between the gray value of the pixel point and the fitting center. In order to accelerate the evolution of the model, the absolute difference of the median value of the pixel inside and outside the curve is introduced to replace the energy coefficient of the inner and outer region of the model, and the efficiency of model segmentation is improved. Finally, it can quickly extract the flame region from the early fire image. A large number of experimental results show that compared with the existing Otsu algorithm and CV model, the CV model with energy weight, the CV model with gradient information and two similar CV models, the improved CV model is used to analyze the early mine fire images. It can achieve better segmentation effect and meet the real-time requirements.
【作者单位】： 南京航空航天大学电子信息工程学院;安徽理工大学煤矿安全高效开采省部共建教育部重点实验室;
【基金】：煤矿安全高效开采省部共建教育部重点实验室开放基金资助项目(JYBSYS2014102)
【分类号】：TD752;TP391.41
[Abstract]:The grayscale values of flame region, flame afterglow and non-flame high gray level interference region in early fire images of underground coal mine are very close, so it is difficult to extract the flame region accurately by using traditional Chan-Vese (CV) model. To solve this problem, an improved CV model is proposed to achieve accurate segmentation of early fire images in coal mines. When calculating the fitting center of the target and background region, the adaptive weight value is introduced to weighted average, and the difference between the pixel gray value and the fitting center is fully considered, and the contribution of the point to the fitting center is determined according to the difference between the gray value of the pixel point and the fitting center. In order to accelerate the evolution of the model, the absolute difference of the median value of the pixel inside and outside the curve is introduced to replace the energy coefficient of the inner and outer region of the model, and the efficiency of model segmentation is improved. Finally, it can quickly extract the flame region from the early fire image. A large number of experimental results show that compared with the existing Otsu algorithm and CV model, the CV model with energy weight, the CV model with gradient information and two similar CV models, the improved CV model is used to analyze the early mine fire images. It can achieve better segmentation effect and meet the real-time requirements.
【作者单位】： 南京航空航天大学电子信息工程学院;安徽理工大学煤矿安全高效开采省部共建教育部重点实验室;
【基金】：煤矿安全高效开采省部共建教育部重点实验室开放基金资助项目(JYBSYS2014102)
【分类号】：TD752;TP391.41
【参考文献】
相关期刊论文 前10条
1 商蕊;方俊;王静舞;;高原低压环境小尺寸池火图像特征[J];燃烧科学与技术;2016年04期
2 吴诗Zs;吴一全;周建江;孟天亮;;利用倒数灰度熵和改进Chan-Vese模型进行SAR河流图像分割[J];测绘学报;2015年11期
3 王相海;王金玲;方玲玲;;灰度差能量函数引导的图像分割自适应C-V模型[J];模式识别与人工智能;2015年03期
4 孙传猛;曹树刚;李勇;;基于LBF改进模型的煤岩细观结构研究[J];煤炭学报;2015年02期
5 宋昱;吴一全;毕硕本;;边缘修正CV模型的卫星遥感云图分割方法[J];光学学报;2014年09期
6 韩郁，
本文编号：2235241
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2235241.html
