Based on extracting the features of grinding crack images of automotive gears, the method of support vector machine is used to improve the recognition accuracy and recognition time of grinding crack images of automotive gears. Support Vector Machine (SVM) is a classification algorithm with learning capabilities that can solve multiple linear and nonlinear parameter problems. Its basic operating principle is related to hyperplane spacing. Support vector machines (SVM) use hyperplane intervals to segment feature samples for the purpose of feature classification of grinding crack images. In the process of feature sample segmentation, the weight vectors of each feature category are calculated to avoid category confusion or classification errors in two-dimensional space. The feature segmentation of grinding crack images using support vector machines is shown in the figure.
The feature weight vectors of grinding crack images of automotive gears can be iteratively solved using particle swarm optimization algorithms. Let the feature decision variable of the grinding crack image of an automotive gear be f, and when the target weight dimension is f (k), the feature decision space is:
According to the particle swarm optimization algorithm in the k-dimensional decision space, the feature vector weights of the grinding crack images of automotive gears can be obtained.
The feature optimal weight vector of the grinding crack image of the automobile gear is input into the support vector machine to obtain the feature classification matrix of the support vector machine on the grinding crack image of the automobile gear, thereby completing the grinding crack image recognition. The classification function formula is as follows:
Where: C is the total number of feature categories; α^ I is the scale coefficient; β^ J is the characteristic classification value of the grinding crack image of the automobile gear; S is the number of iterations of the decision space in the particle swarm optimization algorithm. According to the classification matrix results, it is possible to recognize the grinding crack images of automotive gears.