改进的光流法及其在云爆弹研究中的应用  

Improved optical flow method and its application in the study on FAE

作　　者：周维[1] 鲍远律[1] 於俊[1] 邓明之[1] 

机构地区：[1]中国科学技术大学自动化系GPS实验室,安徽合肥230027

出　　处：《光电工程》2006年第2期9-11,共3页Opto-Electronic Engineering

基　　金：国家自然科学基金(60272040)

摘　　要：在Horn-Schunck和N.Ohta的传统光流法的基础上,提出了用于抑制光线影响的改进的光流法,并将其用在云爆弹的爆炸过程分析研究中,得到云爆弹爆炸云雾扩散速度这一重要参数指标。改进的方法运用自适应高斯核平滑滤波器,对所拍摄的图像序列进行平滑滤波处理,以平滑图像和降低随机噪声对图像处理的干扰。利用rgb彩色模型在白光照射下的不变性原理,将传统彩色图像光流法中的RGB彩色模型转换为rgb彩色模型,得到了一种新的可抑制光线变化影响光流场算法,以减少因光线变化而造成对目标的误检测的影响。以云爆弹为研究实例,采用改进的方法计算云爆弹爆炸云雾图片中各点光流,并提取了扩散云雾平均速度信息。实验所得图片中平均光流大小约2.5pixel/ms,经过参数换算得出云雾扩散速度为109.78m/s,取得了良好的效果。Based on the conventional method of optical flow estimation proposed by Horn-Schunck and N.Ohta, an improved method is presented and applied in the study on Fuel dispenser Air Explosive （FAE） in this paper. By this improved method we can acquire the velocity of the diffusing cloud in explosion. Firstly, images are preprocessed by Gaussian kernel smooth filter in order to suppress the random noise on the images and smooth the images; Secondly, due to the invariability ofrgb color model, we convert RGB model to rgb model to restrain the influence of the changing illumination; Finally, the improved optical flow method based on rgb model is applied in the estimation of the optical flow and the velocity of the diffusing cloud in explosion. Color image sequence of explosive course of FAE is acquired by 1000 frames/s camera, and velocity of diffusing cloud is calculated by means of the two mentioned methods. The better result of 2.5pixel/ms （about 109.78m/s） is obtained by the improved method, whereas some motionless area is detected mistakenly with conventional one.

关 键 词：光流 云爆弹 高斯核平滑滤波器 彩色模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]