Title: Neural network classifier based on genetic algorithm image segmentation of subject robot optimisation system

Authors: Hongbo Ji; Mingyue Wang; Mingwei Sun; Qiang Liu

Addresses: School of Computer Science, Tonghua Normal University, Tonghua 134001, Jilin, China ' College of Science and Engineering, College of Humanities & Sciences of Northeast Normal University, Changchun 130000, Jilin, China ' School of Computer Science, Tonghua Normal University, Tonghua 134001, Jilin, China; College of Computer Science and Technology, Jilin University, Changchun 130000, Jilin, China ' State Key Laboratory of Automotive Simulation and Control, Jilin University, Changchun 130022, Jilin, China; School of Automotive Engineering, Jilin Engineering Normal University, Changchun 130052, Jilin, China

Abstract: Robot optimisation system is a kind of complex, nonlinear, strong coupling system with serious uncertainty. The effect of image segmentation has become an important index to judge the merits of many algorithms. The purpose of this study is to explore the effect of neural network based on genetic algorithm on image segmentation in the optimisation system of classifier subject robot. The method used in this study is to calculate the pre trained VGGl6 NET model as the pre training model through the framework of genetic algorithm. The resolution of the training picture used is 640 * 480, the learning rate is 10⁻⁵, the value of batch size is l, the number of iterations is set to 12,000 and then the trained model is used to detect the image. The results show that the average error of group B of SNN trained by BP algorithm is 11.62%, the SNN trained by SGA has reduced the result to 9.75% and the error reduced to 7.75% by the genetic algorithm in this study. Moreover, genetic algorithm is better in feature point extraction, and the detection rate reaches 94.62%, which is higher than 77.53% and 88.74% of other methods. The missing rate of this study is only 3.04%, far lower than 12.49% and 7.36%. The conclusion is that our genetic algorithm has obvious advantages, small error, high efficiency and applicability. The neural network based on genetic algorithm in this study has a certain value in image segmentation technology.

Keywords: genetic algorithm; neural network classifier; robot optimisation system; image segmentation; feature point extraction.

DOI: 10.1504/IJGUC.2021.119573

International Journal of Grid and Utility Computing, 2021 Vol.12 No.4, pp.369 - 379

Received: 05 Jul 2020
Accepted: 21 Aug 2020
Published online: 09 Dec 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article