An Overview of Image Segmentation Based on Pulse-Coupled Neural Network
- PDF / 1,779,509 Bytes
- 17 Pages / 595.276 x 790.866 pts Page_size
- 41 Downloads / 238 Views
ORIGINAL PAPER
An Overview of Image Segmentation Based on Pulse‑Coupled Neural Network Jing Lian1,2 · Zhen Yang1,2 · Jizhao Liu3 · Wenhao Sun1,2 · Li Zheng1 · Xiaogang Du1 · Zetong Yi1 · Bin Shi4 · Yide Ma2 Received: 15 April 2019 / Accepted: 18 November 2019 © CIMNE, Barcelona, Spain 2019
Abstract Recent many researchers focus on image segmentation methods due to the rapid development of artificial intelligence technology. Hereinto, pulse-coupled neural network (PCNN) has a great potential based on the properties of neuronal activities. This paper elaborates internal behaviors of the PCNN to exhibit its image segmentation abilities. There are three significant parts: dynamic properties, parameter setting and complex PCNN. Further, we systematically provide the related segmentation contents of the PCNN, and hope to help researchers to understand suitable segmentation applications of PCNN models. Many corresponding examples are also used to exhibit PCNN segmentation effects.
1 Introduction Image segmentation is regarded as one of the most important issues for image processing, and its main characteristic is that an image is divided into a certain number of regions according to the static and dynamic properties of the images. When image segmentation results are given by appropriate segmentation algorithms, we can further extract related features, and conduct the identification and classification of objects. In the past decades, Eckhorn et al.’s [1–4] bio-inspired neural network based on cat visual cortex, can synchronously release pulses for similar neuron inputs. Johnson et al. [5–10] developed the above model and proposed a pulse coupled neural network (PCNN). Subsequently,
* Jing Lian [email protected] 1
School of Electronic and Information Engineering, Lanzhou Jiaotong University, No. 88 West Anning Road, Lanzhou 730070, Gansu, People’s Republic of China
2
School of Information Science and Engineering, Lanzhou University, No. 222 South Tianshui Road, Lanzhou 730000, Gansu, People’s Republic of China
3
School of Data Science and Computer Science, Sun Yat-sen University, No. 132 East Outer Ring Road, Guangzhou 510275, Guangdong, People’s Republic of China
4
Equipment Management Department, Gansu Provincial Hospital, No. 204 West Donggang Road, Lanzhou 730000, Gansu, People’s Republic of China
Ranganath and Kinser et al. [11–15] presented modified PCNN models and further exploited image processing capacities of the PCNN. PCNN has broad applications in image processing field, such as image fusion, image segmentation, image denoising, image enhancement, feature extraction. In recent years, PCNN have significant potentials for evolving image segmentation algorithms, due to its synchronous dynamic properties of the neuronal activity, including synchronous pulse release, capture behavior, nonlinear modulation and automatic wave. Further, in contrast to other prevalent segmentation methods, PCNN has low computational complexity and high segmentation accuracy, which becomes quite suitable for image
Data Loading...
