基于改进的SIFT算法的阀体气密性检测方法

Valve Body Airtightness Detection Method Based on Improved SIFT Algorithm

为解决目前阀体气密性检测存在效率低、人为因素影响大等问题,提出一种基于改进的SIFT算法的阀体气密性检测方法。首先采用差影算法对现场采集到的图像进行预处理;然后基于SIFT算法,结合改进的SUSAN算子及仿射变换对图像进行匹配拼接;最后循环遍历形状元素得到气泡的真实参数,从而判断阀体是否合格。实验表明,该方法提高了气泡检测效率且检测准确率达98%,适用于同类阀体气密性检测。

In order to solve the problem of low efficiency and large influence of human factors on the airtightness testing of the valve body,a method based on improved SIFT to detect the airtightness of the valve body is proposed.First of all,the algorithm of difference shadow is used to preprocess the collected images;Then based on the SIFT algorithm,combined with improved SUSAN operator,affine transformation,the image mosaic is matched;Finally,the traversal of the shape elements is repeated to obtain the real parameters of the air bubble,so as to judge whether the valve is qualified.The experiments show that the method improves the bubble detection efficiency,and the detection accuracy reaches 98%,which is suitable for testing the airtightness of the same of valves.