Abstract
With the continuous development of manufacturing and the advancement of science and technology, products such as precision workpieces have increasingly stringent requirements for manufacturing accuracy. Traditional manual measurement is becoming less capable of meeting the demands of high-precision production, because it requires direct contact with the workpiece and is susceptible to measurement errors caused by subjective factors on the part of operators, which also reduces measurement efficiency. At present, high-precision, intelligent, non-contact, and efficient vision-based dimensional measurement systems have attracted increasing attention. To address the difficulties in measuring the dimensions of circular precision workpieces and the relatively high cost of inspection, this paper proposes a fast and accurate detection algorithm based on a regional gray-level model. A machine-vision-based method for dimensional inspection of circular precision workpieces is investigated. First, image preprocessing is completed through morphological processing. Then, OTSU adaptive thresholding is introduced into the region gray-level model algorithm to improve detection accuracy. Subsequently, defect processing is performed using an annular region defined by the Canny operator. Finally, the circular diameter is fitted using the least squares method. The proposed method has high computational speed and high detection accuracy, and it satisfies the accuracy requirements for workpiece inspection.