基于计算机图像分析的油液污染度测试方法研究

Oil Contamination Test Method based on Computer Image Analysis

传统检测方法存在抗干扰性弱、检测误差大的问题,为了解决该问题,提出了基于计算机图像分析的油液污染度测试方法。根据图像获取原理重现油液污染颗粒含量,使用梯度向量来表示图像灰度变化,采用神经网络检测算法对图像边缘进行检测分析,利用学习向量量化神经网络对颗粒识别,并在识别图像文件获取颗粒污染度含量,并设计污染度评定流程。通过实验验证结果可知,该测试方法抗干扰性强,误差始终没超过20%,检测结果与传统方法相比较为准确。

The traditional detection method has the problems of weak anti-interference and large detection error. In order to solve this problem,a method of oil contamination measurement based on computer image analysis is proposed. According to the principle of image acquisition,this paper reproduce the particle content of oil pollution,to represent the image gray change using gradient vector,using neural network detection algorithm of image edge detection and analysis,using learning vector quantization neural network of particle recognition,and obtain the contamination content in recognition of the image file,and design process of the assessment of pollution. The experimental results show that the test method has strong anti-interference. The error is not more than 20%,and the detection result is more accurate than the traditional method.