The Image Noise Suppression Method And Feature Matching Method For Large Forgings Based On Machine Vision

Due to the extreme complex working environment, nuclear power rotor, marine crankshaft and other large spare parts require high performance. Forging process is usually used. The processing capacity is the embodiment of the state's extreme manufacturing capacity and manufacturing level, also, an important guarantee for the national economy and national defense security. The existence of the manufacturing technology of large forging production is not mature, such as the high cost and the large energy consumption. Therefore, the accurate measurement of geometric parameters of high temperature forging is necessary to reduce the waste rate, the cost, the energy consumption. Laser-aided vision measurement with advantages of non contact measurement, safety, convenience and high efficiency can solve the problem of accurate and safe measurement of the geometric parameters of the forging in the complex environment.In laser-aided vision measurement system, effective information extraction, information noise suppression and feature matching are the core parts of image processing and the key link in the overall measurement accuracy. However, because of the complexity of the forging measurement environment, the image information and noise are very difficult to separate and it is difficult to make full use of the effective information to measure. At the same time, problems existed in the application of feature matching method. Existing problems will introduce errors to the measurement. In this paper, noise suppression and feature matching are studied:(1) Gray feature information of the centre points of the light stripe along the light direction is discussed. Laboratory experiments show that the parameters of the distribution are in accord with the physical meaning, and more in line with the actual distribution than quadratic curve and gauss curve.(2) A method for judging the credibility of the centre points is proposed. The ratio of the actual gray energy and the theory gray energy is used to judging whether the centre points are accurate extraction. The actual grey energy is on the image, and the theory grey energy is obtained by fitting. In the experiments conducted, the relative error of laboratory measurements using the proposed method reduced from 1.84%to 0.33%. Experiment in the field of forging shows that the method has a good effect on removing the noise from the forging scene, and the precision of forging measurement is greatly improved.(3) According to the condition of the measurement field, constraint conditions for image feature matching which is suitable for forging measurement is discussed. A feature matching method based on the pole line constraint is proposed.