DISCUSSING ACCURACY IN AN AUTOMATIC MEASUREMENT SYSTEM USING COMPUTER VISION TECHNIQUES

A Computer Vision system extracts and analyzes image features in order to recognize and/or measure it. A Computer Vision system for inspection has received considerable industry attention, in the last years. This paper uses some Computer Vision techniques to measure parts and discusses common difficulties of automated inspection. The parts conformity analysis using a non-contact measurement system has been adopted specially to small objects, where accurate instruments are not so simple to be used. In these cases Coordinate Measuring Machine (CMM) can be used. However, in many times it is impossible to move the object to the CMM. In this case an image automated inspection should be a good alternative. The proposed system consists in a CCD camera positioned in the upper viewer of the object. Some considerations must be pointed out: (1) the object should be thin, to reduce the height influence in the measurement; (2) a standard gage must be used to convert pixels units to millimeters units. We present the algorithms developed to make dimensional measurements in industrial parts. A rectangle object with three holes is analyzed. The obtained results are compared with convencional instruments measurements. We highlight the difficults concerned to image automated inspection, discussing accuracy and the relationships among images and measurements.