A Liquid Level Detection Method Based On X-ray Imaging System

Liquid level detection with the use of machine vision and image processing technologies is accurate, reliable and non-destructive. This study is of canned beverage products, focuses on X-ray liquid level images of metal cans, works out the liquid level detection based on machine vision and image processing.In this paper, the principle and parameters of X-ray imaging system are elaborated, the characteristics of X-ray images of canned beverage is analysed, the relationship between gray level and X-ray tube voltage or tube current is discussed by experiments, and also the appropriate imaging parameters is selected. On the basis of analysing the source and characteristics of X-ray image noise, a denoising filter based on Pulse Coupled Neural Network is proposed. Experiments have showed that the filter makes up for disadvantages of classical filtering method.Due to the inefficiency of the traditional mark method in liquid level measurement by image, a mark method based on the rule'ratio of actual distance corresponds to ratio of distance in pixel'is presented according to the physical characteristics of cans. A new algorithm on X-ray liquid level detection based on Canny method is proposed, it include: improved Canny edge detection, edge connection of liquid level based on curve fitting, information extraction by statistical method, and liquid level calculation. Based on Hough Transformation, an automatic calibration method for tilted cans is proposed to improve the accuracy of the liquid level detection algorithm.This paper gives the simulation test on the above image processing and liquid level detection algorithms in Matlab 7.0 environment. A liquid level detection software on X-ray image is developed in Visual C++, embedded the algorithms above, it performs the detection in experimental system.