A new image segmentation method based on the ICSO-ISPCNN model
- PDF / 3,039,892 Bytes
- 24 Pages / 439.37 x 666.142 pts Page_size
- 91 Downloads / 185 Views
A new image segmentation method based on the ICSO-ISPCNN model Jianhui Liang 1,2 & Lifang Wang 1
& Miao Ma
3
Received: 13 December 2018 / Revised: 24 October 2019 / Accepted: 20 December 2019 # Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
To address the issue of parameter settings in a pulse coupled neural network (PCNN), we propose a new image segmentation method based on the improved chicken swarm optimization algorithm and improved simplified PCNN (ICSO-ISPCNN) model. First, we improved a simplified PCNN model by modifying the dynamic threshold function and meanwhile improved the chicken swarm optimization (CSO) algorithm by introducing the survival of the fittest mechanism. Then, a product cross entropy is utilized as the fitness function of the ICSO algorithm, and the parameter values of the ISPCNN model are determined through the effective teamwork of roosters, hens, and chicks in the chicken swarm. Finally, we can achieve the automatic image segmentation via the ISPCNN model, which has the best parameter values. The detailed experiments indicate that our method has more superior performance in terms of convergence and segmentation accuracy than methods based on the genetic algorithm and ant colony optimization algorithm. Keywords Chicken swarm optimization algorithm . Image segmentation . Pulse coupled neural network
1 Introduction The pulse coupled neural network (PCNN) is a mathematical model that is proposed based on the characteristics of neuronal activity in the cerebral cortex of cats and other mammals. It is a * Lifang Wang [email protected] * Miao Ma [email protected]
1
School of Computer Science, Northwestern Polytechnical University, Xi’an 710072, China
2
College of Applied Science and Technology, Hainan University, Danzhou 571737, China
3
School of Computer Science, Shaanxi Normal University, Xi’an 710062, China
Multimedia Tools and Applications
single-layer network model that can achieve pattern recognition and image segmentation without any learning and training. Therefore, it is ideal for the real-time image processing environment, and it has attracted extensive attention of researchers at home and abroad [4, 8, 13]. Image segmentation is a process of dividing an image into regions with different characteristics, it plays a vital role in the subsequent recognition and understanding [5, 10–12]. At present, there are thousands of segmentation methods, including threshold-based segmentation, region-based segmentation, edge detection-based segmentation, specific theory-based segmentation and content-based segmentation, such as semantic segmentation,etc. This kind of segmentation method makes full use of the content information of an image, such as image semantics, to segment an image, therefore, it is more suitable for image segmentation of complex scenes than traditional methods. But, a large number of labeled images need to be trained as samples [1, 32, 33, 37]. The PCNN model has been successfully applied to image segmentation since its synchronous
Data Loading...
