Holes Position Testing Method Research For Plate Parts Based On Photogrammetry

With the rapid development of the industry, the demand of the precision and accuracy for machining and testing is increasing. The existing industrial products processed with holes of various shapes and the location precision of these holes will affect subsequent use of them. Therefore, we need to detect the machining accuracy of the holes. Image measuring, as a measure of non-contact measurement method, is being widely used in detection of industrial products due to its high performance and low price. This paper is to study the hole position detecting method based on image.Due to the images captured by camera are color images and will contain noise, therefore, pretreatment including converting color images to grayscale images and denoising are required. This paper introduces the basic principles of converting color images to grayscale images and two commonly used spatial filters: median mean filtering. In this paper, relevant experiments of median and mean filtering are done. Due to the method of edge detection is the base of two center location methods of circle: circle’s center location based on fitting algorithm and circle’s center location based on the geometric feature of circle, this paper makes a detailed introduce of some commonly used methods of edge detection. Canny operator is selected in this paper for its high positioning precision and single-edge test results according to experiments.On the bases of image processing and edge detection, this paper introduces two circle detection methods that based on image and makes a simple comparison and analysis of these two methods, concluding that the circle’s center position method based on fitting algorithm has a relatively higher accuracy. This paper proposed a new method of testing the processing accuracy of irregular shapes. The main principles of this method are as follows: firstly, taking photo of the plate parts which is to detect and getting the region of irregular shape through template matching. Secondly, getting the edge points collection of the irregular shape. Thirdly, getting perspective transformation parameters between actual image and CAD model and making conversion of the edge point collection of irregular shape using perspective transformation parameters. Lastly, making comparison between the edge points collection we got in the previous step and CAD model using the data analysis method based on free curve. Thus, we can get the processing precision of irregular shape.Finally, this paper realizes the methods above through programming. Through the relevant experimental analysis we can conclude that the accuracy of hole center location based on fitting algorithm is able to reach 0.0006 pixel theoretically and 0.1 pixel for actual shooting image, achieving sub-pixel positioning. The method proposed in this paper for irregular shape detecting can reach 0.2 mm and it does not require object to test perpendicular to photography-axis strictly. Therefore,it can meet the detection requirements for irregular shapes of most industrial products.