Research On The Straightness Detection Of Axis Parts With Image Measurement

In the field of mechanical manufacturing , rotational parts (such as axis, plate, covered, hole and so on) account for a large proportion, and those parts'position tolerances (such as straightness) have significant impacts on their performance. It has a direct impact on the surface's attribute, the accuracy of the rotary position's surface, thereby affects the entire rotation accuracy of the machines and some other important parameters. But the major detection methods usually are traditional contact-detection ones, these methods have many problems and errors.For overcoming the shortcomings of traditional methods, and adapting to the new situation and the need of new development, this subject researches a new straightness detection method that bases on image measurement, and develops the related software system. The main work of this subject is to research following several contents.1. Measure principle and system constructionsThe hardware system with video image processing technology consists of the light system, CCD camera, PCI card, computer, and output equipments. The software system consists of image show, calibration the input data, output the data of the axis, the output of straightness data, the output of the coordinate and graphics. Work principle is: firstly, collect the axis images by CCD camera, transport the picture into the computer by PCI card, then get the edge of the axis by digital image processing technology, finally, get the straightness though the filter und centerline fitting. For the problem of the calibration, the axis edge in picture is proportional to the actual one, and the ratio is a constant βdetermined by CCD camera magnified ratio.2. Digital image processing and measuringThe basic characteristic of the image is the edge. In the image processing, edge is important to both image division and measurement. Meanwhile, edge measurement is the key technology in non-contact measurement. Its results have directly impact on precision. This subject uses both rough edge detection and precise edge detection. First, search the rough axis edge by phasing in dispersion, the precision of location attains the pixel level. For attaining the higher precision in searching the rough edge, it uses the subpixel arithmetic---one dimension gray moment operators. It breaks the restrict of the CCD camera physics resolution and makes the location precision attaining the subpixel level.3. Straightness error detectionFor one frame image, firstly, we should fit the axis's centerline, and then use the integrated evaluating method by two-endpoint and least square method to solve the straightness error in the image, then use the system ratioβto get the geometry value of the axis straightness. According to this method, it can be calculated the axis straightness error in different degree, and choose the maximum to be this axis straightness error t . In this way we complete the whole process of the axis straightness detection.4. The factors of influencing the measuring precisionIn this subject, we find the factors what influence the measuring precision by experiment. Though measuring precision of the image measurement system depends mainly on resolution ratio of the digital camera, we must consider many factors which affect the precision in order to improve the precision, the main factors are as follows: lighting system error, the installation error of CCD and lens, error of geometrical aberrance of imaging system, current noise error, quantization error, the error that is involved by environment and the influence of the axis. By analyzing the factors of errors, we choose the best experiment condition, and in large extent to reduce the system error and improve the measuring precision.5. ConclusionThe key problem of the video image measuring system is the system precision. Higher precision means better measuring result, higher accuracy. In order to improve the system veracity, we have done many experiments. Through comparing different experiment conditions and analyzing experiment results, we can choose the best experiment to be the condition of the following study. In order to improve the precision, we use the method of mean filter and axis centerline fitting using symmetry method.On this basis, using the integrated evaluating method by two-endpoint and least square method to solve the straightness error. Experiment results show that this method on video image measurement of straightness detection is correct and feasible.