Research On Image Recognition Method Based On Neural Network

With the spread of computer technology,digital image technology has penetrated into every field of life and production,the transmission and storage of digital image technology is already very mature.As the core of digital image technology,the image recognition technology has been a hot and difficult point in domestic and overseas research.The basic idea of this technique is using the computer to classify the different scenes in the obtained images automatically instead of the traditional tasks of manual completion.Due to the limitations of traditional image recognition methods,there are still problems such as narrow scope of application,poor flexibility and insufficient robustness in the application process.The neural network provides a new and effective method to solve the problem of image recognition because it has less requirement of prior knowledge and can complete the complex division of feature space.Based on the research of neural network and its role in image recognition,this paper proposes an image recognition model based on BP neural network.Aimed at the shortcoming of BP algorithm is easy to fall into local minimum value,the structure of the BP neural network and BP algorithm to do the research on the basis of full analysis,this paper proposes a driving factor of the improved algorithm.On the basis of predecessors' research,this paper constructs a set of moments to distinguish different image targets.The image recognition experiment is carried out by using the invariant moment feature of the extracted image sample.The experimental results show that the model is effective and feasible and has a good recognition rate.