Experiences about X-ray based machine visionX-ray based machine vision has been one research topic in Optical Measurement Laboratory of Kemi-Tornio University of Applied Sciences in year 2005. The main features of digital x-ray imaging have been studied from a perspective of making dimensional measurements and object classification. Among other things a simulation tool for attenuation of x-rays with matters and technology for bio fuel quality control has been developed. Introduction The Optical measurement laboratory of Kemi-Tornio University of Applied Sciences has about eight years of experience about visible light machine vision technology. During those years many kind of a machine vision based technologies have been developed to industrial applications. For example a glass measurement and identification system for glass tempering optimisation and a machine vision controlled laser cutting system for plastic parts trimming. From the beginning of year 2005 a new area of machine vision technology for Optical measurement laboratory has been started to study. Machine vision expertise has been extended to non-visible light machine vision technologies. X-ray and thermal camera imaging have been focus areas in mainly Tekes-funded research project called Development of x-ray and thermal camera imaging to accurate dimensional measurements and object classification (Finnish abbreviation R-IP) during the last year. This article gives a short introduction to x-ray based machine vision technology, presents an x-ray attenuation simulation tool and describes an x-ray based machine vision application developed and researched in the project. X-ray based machine vision X-rays are a form of electromagnetic radiation with a wavelength in the range of 10 to 0.1 nanometres. X-rays can be produced with x-ray tubes designed for x-ray imaging applications. Radiation coming from the tube can be detected by using digital x-ray detectors. The detectors can be linear or matrix shaped, the linear detectors are commonly designed for imaging of moving objects. X-ray based machine vision is commonly based on transillumination principle, where the target object between x-ray tube and detector attenuates radiation based on physical characteristics and thickness of the target object (Figure 1.). The radiation, which is left after the target object can be detected with the x-ray detector. A digital image acquired from the detector to some processing unit (e. g. PC computer) can be processed by using same kind of image processing methods as in normal visible light machine vision systems.  The best known application areas for digital x-ray imaging might be the medical and security applications. The x-ray based machine vision is widely used also in industrial applications. For example the food and electronic industries are using x-ray based machine vision in their quality control. Detection of unwanted items inside the food products or quality inspection of BGA components are some examples of those sectors of industrial x-ray applications. The major benefit of x-ray based machine vision compared to visible light machine vision is naturally the fact that x-ray system can see inside the objects. It is quite probable that the use of x-ray based machine vision technology will increase significantly in the future. The development of the x-ray system components is rapid at the moment. The cost efficiency of the systems, which has been the most limiting factor of applying the x-ray technology more widely, is improving quickly all the time. Based on that development trend, it is even possible that the x-ray systems will substitute some of the traditional metal detectors in security and industrial applications in the future. X-ray system can detect and also classify (position, shape, thickness etc.) all the metal objects and other materials of interests in those applications. That is a benefit compared to traditional metal detectors. Simulation of x-ray attenuation A simulation of x-ray attenuation was one of the goals in the R-IP project. The developed x-ray attenuation simulation is a tool for rough designing of x-ray based machine vision systems. The simulation can be used to find out the relative amount of radiation left after known materials with known thicknesses. That information is valuable when estimating if there is contrast enough for target object detection inside known materials and in that way possibilities to have a good quality image. The simulation also can be used to find out the optimal tube voltage based on the known parameters of the object materials. The attenuation of the x-ray going through the target object can be calculated by using the formulas of radiation physics when the thickness, density and attenuation coefficients of the target object are known. The attenuation varies in different radiation energies. A discrete data about attenuation coefficients as a function radiation energy about different matters can be found e.g. from NIST tables [1]. When the radiation spectrum of the x-ray source with certain tube voltage is known, and the x-ray attenuation is calculated, a relative amount of intensity of radiation left after the target object is a result of the simulation (Figure 2.).  X-ray based machine vision application One case example in the R-IP project was to detect foreign matters from the bio fuel. The bio fuel in this case was mainly peat. The goal was to detect rocks, metals, iced peat cubes and other possible impurities inside the peat conveyor. The x-ray based machine vision system was developed to detect these impurities. The possible impurities can cause a break in the bio fuel handling process if they are not detected and removed early enough. The peat conveyor is a kind of box shaped steel tube, and the peat is moving inside the closed conveyor. The first step was to perform offline tests for evaluating x-ray based machine vision functionality in practice. A sample of the bio fuel and the impurities was transilluminated by using an x-ray system owned by a co-operation company of the Optical measurement laboratory. These offline tests showed (Figure 3.) that the x-ray system can detect the demanded impurities inside the bio fuel in certain conditions. The x-ray attenuation simulation tool was used to decide that a line shaped holes (due to line x-ray detector) to bottom and front steel covers of the conveyor must be made to achieve a pass enough for the radiation for successful transillumination of the peat mass. In this case the x-rays must go through 18 mm steel if the conveyor covers are not modified. Based on the simulation, only 0.14 percent of radiation by using 140 kV tube voltage is left after 18 mm steel. This was not enough for achieving acceptable contrast for detecting the impurities from the peat. Thus, the line shaped holes were machined to the conveyor top and bottom covers and the holes were filled with plastic materials.  The developed technology will be tested in real environment in spring 2006. The x-ray based machine vision system will be installed to peat conveyor, and the image data is collected from approximately a month period. Developed image processing methods will be applied to collected image data. If the whole technology is working appropriately, an on-line impurity detection system to bio fuel can be developed. Also new application areas for the technology are looked for in the future. Discussion The x-ray based machine vision is extremely interesting area of machine vision technology. The idea of seeing what human cannot see is fascinating and it opens new possibilities to machine vision applications. Also the strong belief in mushrooming of x-ray applications and the fact that the x-ray based machine vision is not so much researched area than a visible light machine vision gives the bases to carry on the research of the x-ray based machine vision in the future. A new research and development project is under planning. Technology to moving target objects and cost efficient systems in industrial x-ray applications will be developed. Also the multi-energy methods in x-ray transillumination systems are one technical focus area in the future project. References [1] http://physics.nist.gov/PhysRefData/XrayMassCo-ef/cover.html
|
