Tutorial: line scan cameras
Tutorial: line scan cameras
复制标题
教程:线扫描相机
DOI:
10.1108/eb007797
复制
发表时间:
1989
期刊:
影响因子:
--
通讯作者:
C. Loughlin
中科院分区:
文献类型:
--
作者:
C. Loughlin
We shall look first at the procedures necessary to generate a two-dimensional picture from successive line scans. Once this picture has been obtained, then the same, or similar techniques to those used for area array image processing and inspection can be applied, and these will be covered in a future article. The line scan camera takes repetitive scans of the scene in front of it and these are sent to a scrolling field store and computer system that analyses the two-dimensional scene that builds up. The two dimensions of the area array being replaced by the single dimension of the line scan camera and the extra dimension of time. If the product is moving past the camera at a continuous speed there are numerous advantages in using a line scan camera as opposed to an area array. The first and most obvious is the higher resolution of the image that can be obtained. Resolu tions of 512 by 512 are now generally available in area array cameras (Fig. 2), with the largest commercial array currently known being Kodak's Megaplus camera with a resolution of 1320 by 1035 pixels. By contrast line scan cameras with up to 2048 elements are now in common use and 4096 element arrays are readily available. If a line scan camera is used to monitor a continuously moving produc tion line, then the image that is caught is essentially the width of the production line and infinitely long, whereas in the case of an area array camera, rect angular snap shots of the production line are obtained, which will either overlap one another or have missing gaps between successive frames (Fig. 3). This can cause confusion in applica tions such as product counting when parts can be missed or counted twice. Further, the exposure time (i.e. the time between one scan and the next) for a line scan array is typically less than 1 ms, whereas area arrays generally operate at 16 or 20 ms (60/50 Hz) integration times. Long exposure times result in a blurring of the image and inaccuracies in dimensional measurements, as well as making the task of feature recognition more difficult (Fig. 4). It would be wrong to give the impression that line scan cameras are superior to area arrays in all respects as there are numerous applications for which area array cameras are ideally suited, and many area cameras are now available with very high speed elec tronic 'shutters' or higher frame rates than the 50/60 Hz video standards that help overcome some of the difficulties outlined above. We can illustrate the techniques of 2-D image capture in the following worked example that is based on a real application. A line scan camera with 2048 elements and a minimum cycle time of 200 microseconds per scan (5000 scans/s) is used to monitor a production line that transports products that are 200 mm across and moving at a speed of 1 m every 5 s (200 mm/s). The 2048 elements of the line scan camera enable us to achieve an inspec tion resolution across the product of approximately 0.1 mm. If we also want to inspect in the direction of movement at the same resolution, then we will need to sample 2048 scans within the 200 mm that the conveyor belt moves in the second. This requires a scan time of approximately 500 microseconds which is within the specification of the line scan camera. If it is known that the product is moving at a constant and even velocity, then it is sufficient for the scans to be triggered by a programmable clock within the camera (Fig. 5a). However, it is often the case that the product will be mounted on a conveyor belt the speed of which will be subject to both long term drifts and short term judders. In these cases it is necessary to have an encoder mounted onto the drive shaft of the conveyor system, or in some other way linked to the belt's movements, such that it sends a line trigger pulse to the camera for every 0.1 mm of actual conveyor belt movement (Fig. 5b).