基于图像处理的试管位置实时检测

Real-Time Detection of Tube Position Based on Image Processing

摘要: 为解决人工识别试管位置效率低、易出错问题，利用图像处理技术，设计一种对于医学冷藏盒上的试管位置自动实时检测识别的方法。通过USB摄像头读取视频流，USB摄像头要进行相机标定去畸变，利用OpenCV读取获得的每一帧图像，并保留第一帧图像。在图像上确定检测的冷藏盒的ROI矩形范围、冷藏盒上每个试管格的四个顶点以及四个顶点的中心点，每帧图像与第一帧图像做差分运算后转换为灰度图，设定阈值并根据这个中心点位置索引获得试管在冷藏盒上准确的相对坐标位置。测试结果表明，该方法可以有效和实时地完成试管位置的检测识别，准确率几乎可达100%，有较大实用价值。

Abstract: In order to solve the problem of low efficiency and error-prone manual identification of test tube positions, a method for automatic real-time detection and identification of test tube positions on medical refrigerator boxes was designed by using image processing technology. The video stream is read through the USB camera. The USB camera needs to be calibrated and de-distorted, and each frame of image obtained is read by OpenCV, and the first frame of image is retained. Determine the ROI rectangular range of the detected refrigerated box, the four vertices of each test tube grid on the refrigerated box and the center point of the four vertices on the image, and convert each frame image and the first frame image into a grayscale image after differential operation. Set the threshold and obtain the exact relative coordinate position of the test tube on the refrigerated box according to this center point position index. The test results show that the method can effectively and real-time complete the detection and recognition of the position of the test tube, and the accuracy rate can reach almost 100%, which has great practical value.