We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Image Recognition of Navel Orange Pest Based on PSO Constrained Optimization and Coupling Local Regularization.
- Authors
Yin'E ZHANG
- Abstract
An image recognition model (GCMPSO-BP) of one gray level co-occurrence matrix and catfish particle swarm optimization neural network for parameter optimization problem of BP neural network is proposed to improve the image recognition of navel orange pest. First, image features have been extracted with gray level co-occurrence matrix, and then the features have been input into BP neural network for study, BP neural parameters have been optimized through particle swarm optimization algorithm, "catfish" effect has been introduced to get over the local optimal defect of particle swarm optimization algorithm and finally specific image database has been adopted to make simulation test for model performance. The simulation results showed that compared with traditional image recognition model, GCMPSO-BP could attain better image recognition performance, which not only improved the image recognition rate and recognition efficiency of navel orange pest but also solved the problem of parameter optimization for BP neural network.
- Subjects
PARTICLE swarm optimization; REGULARIZATION parameter
- Publication
International Journal of Simulation: Systems, Science & Technology, 2016, Vol 17, Issue 40, p1
- ISSN
1473-8031
- Publication type
Article
- DOI
10.5013/IJSSST.a.17.40.07