Image Recognition of Engine Ignition Experiment Based on Convolutional Neural Network

In the engine ignition experiment, the specific instant of the ignition is usually obtained from a large quantity of high-resolution pictures taken with high-speed cameras, which puts forward an urgent request for the rapid image recognition. To address this issue, a picture recognition method based on convolutional neural network (CNN) is described. First, a training data set for the CNN model is made based on the original experimental images. Second, the constructed CNN model is trained to obtain the classification result. Finally, the CNN model is evaluated and optimized for the image recognition of engine ignition. The experimental results show that the method can quickly and accurately recognize the engine ignition.