The selection of a lens requires the consideration of many parameters such as lighting, edge sharpness of the scene, Z axis motion of the scene, and distance from the camera to the scene. The lens provides a projection of the scene into the OpticRAM. This means if the lens is not selected properly or is misadjusted (out of focus, etc.) the -information that the opticRAM sees will not adequatly represent the scene, (for the threshold data slice of the scene will not represent the scene). One will be hard pressed to interpret `'hat the camera is looking at. The choice of a lens in terms of focal length and field of view are directly affected by restrictions which may exist on the working distance of the camera. For example, a room size may restrict the camera from moving back far enough to have the scene in focus or fully captured. The least resolvable element or increment in a measurement system may be the dominant factor, implying that more than one camera may be required in the system. In our system, with a built in threshold sensing technique, the resolution is equivalent to one pixel. The scene resolution is the pixel pitch times the object magnification. Accuracy is the degree of exactness to which the measurement can be made. Under controlled conditions, accuracy can equal the resolution. When measuring the distance between two edges of an image, the accuracy is equivalent to one element per edge under conditions of having a sharp optical image of the object's edge. If lesser accuracy occurs, it is usually due to an unsharp edge, created by poor contrast between the object and the background, or due to dynamic aspects of object movements and integration time. However, by averaging one edge (or edges), the accuracy can be finer than the object resolution. The following example discusses how one would select each component part for the camera and system configuration: A disk is to be measured for its diameter on a translucent conveyor. The conveyor speed is 15 feet per minute. The disk size is .2 inches (with .02 inches of variation) with a height variation of 40 mils. This includes the conveyor thickness variation and vibration. NOTE: The limit tolerance in relation to the nominal size is .02 inches. However, the measurement of the part may require 10 times better resolution than the limit requires, 1% in this case.
In this example we will look at two ways to implement the solution. One solution is using a strobe light while the other solution is to analyze the motion of the part as it relates to the array.
Figure F-4 (GIF 20k) describles the disk on the conveyor.