基于梅林傅里叶变换与感知哈希的表计检测

Meter detection using fourier-merlin transform and perceptual hash

巡检机器人进行表计巡检时,由于室外环境多变,背景复杂,且存在拍摄机器定位不准的情况,针对这一问题提出了一种基于梅林傅里叶变换和感知哈希的电力表计检测方法。首先利用Adaboost分类器对仪表进行粗检测,得到若干个候选仪表区域;然后,利用基于梅林傅里叶变换的感知哈希算法进行目标仪表的精匹配,得到最终仪表区域。实验表明,该方法对光照、噪声等具有一定的鲁棒性,能够适应电力巡检过程中多变的室外环境;且在巡检定位存在一定偏差的情况下保证稳定性和准确性,可满足电力巡检机器人的实际应用需求。

In the process of automatic detection of electricity meters,the outdoor environment is changeable and clutter.In order to solve above mentioned problems,a method based on Merlin Fourier transform and perceptual hashing is proposed in this paper.Firstly,the Adaboost classifier is used to obtain several candidates.Then,the target instrument is matched by the perceptual hashing algorithm based on Merlin Fourier transform.The experiment shows that the proposed method is robust to illumination and noise,and can adapt to the changeable environment.The method is reliable and precise,and can meet the needs of practical application.