Test And Analysis Of Rotation Process Based On Machine Vision

Aimed at testing the time intervals of drill accurately in test and analysis of quality of drill, a new method to test the rotation speed based on Machine Vision is referred to, which involves in the structure of the test system, sampling, time series analysis, error estimating and the carry-out of them by MATLAB language, the advantages of the method and the problem existing etc. According to the theory of Machine Vision ,this method has the very potential value in the fields of studying the working principle of mechanism, measuring the instantaneous torque or the instantaneous power and dialog etc. By this method, sampling needs no special setting of the surrounding, illumination or the technology. But the interval and the evenness of sampling is important to the precision. The frequency of 30Hz was adopted. A Rotation-mark which indicates the location of the rotational parts is designed to make the imaging and the image processing simple and precise enough. It needs to be differentiated easy from the background by the color, and its shape of a rectangle which size is enough by 6mm×1mm. The design of the Rotation-mark and the sticking mainly consider the facility. In order to study this method and to measure the error, a six-levels rotation speed test has been designed and carried out. It covers all the angular speed levels of the drill. The image processing contains the following steps: (1) Compress the non-testing objects in the image frame, such as the base, the shell of the metering mechanism and so on. So colleting the image-samples should avoid to take such non-testing objects. (2) Convert the images to grayscale images. To do so, the Rotation-mark and the background belong to the different domains by the intensity. (3) Image binaryzation. The Rotation-mark is expected to be left only(its value is 0) and the other objects to be suppressed(its value is 1). But in the processed image , there are many micro-size dots which can be considered as noises. According to the gray scale histogram, set the threshold value 0.61. (4) Image filtering. Suppress the noise in the binaryzed image. Set the 模板...