摘要
水面的折射和波动等会使机器视觉采集的图像产生扰动,严重降低视觉图像的质量,从而影响视觉检测结果。提出了一种针对水下扰动视频图像的图像复原方法。该方法基于时空优化的高阶谱分析图像重构方法。由于图像的高阶谱分析需要多维空间,构造算法空间和时间复杂度均较高。首先设计基于图像质量评价的高通滤波,筛选出扭曲较小的待测图像序列;同时通过高斯金字塔模板匹配方法,粗定位水下零件的位置,从空间和时间上同时缩小双谱分析的空间范围;最后通过构造图像序列的双谱,使用双谱重构方法去除扰动影响。实验结果表明,该方法可有效去除水对图像的干扰。
When automatic operation processing using machine vision for measuring under water,disturbance effect caused by refraction and fluctuation of water surface sensitively degrades the video quality,and affecting the detection results. An image restoration method is presented in this paper. The method uses high order spectrum technique. First,a temporal high pass filter is used to get a stabilized but blurry video,which using image quality assessment function. A template match algorithm is also used to local the area of part. Then bispectrum analysis is used to analyze the raw image sequences and recover the phase information of the true object. The experimental results show that the underwater image restoration method can effectively remove the disturbance.
出处
《机电一体化》
2018年第9期50-54,共5页
Mechatronics
关键词
图像复原
双谱分析
高阶谱
模板匹配
image restoration
bispectrum analysis
high order spectrum
template match
