摘要
本文针对在炼钢厂复杂背景下铁水罐定位分割困难这一问题,提出了一种简单有效的解决方法。此方法首先采用帧间差影法初步分割出运动目标,通过对差影图像进行K邻域自适应滤波,消除强脉冲干扰。然后利用帧差图像边缘与灰度图像边缘的强相关性得到目标边缘信息,利用此信息对运动目标进行精确定位,并进一步检验运动目标是否为铁水罐。现场运行结果表明,此方法具有很强的鲁棒性,且算法实现简单,完全满足现场实时性的要求。
In this paper, we present an iron pot detection and segment algorithm for complex background in steel-making plants. First, the difference image of each two neighboring frames is gotten to detect the possible area of moving object. To eliminate the strong impulse disturbance in the difference image, we use the K neighborhoods adaptive filtering. Secondly, using the strong relativity between the edge image of the difference image and of the gray scale image, we can find the real moving object area, and test whether the object is iron pot. The experimental results show that the algorithm is effective and simple.
出处
《微型电脑应用》
2006年第1期13-14,46,共3页
Microcomputer Applications
关键词
差影图像
K邻域自适应滤波
边缘检测
目标分割
difference image neighborhoods adaptive filtering edge detection object segment