An improved image de-noise method based on pulse-coupled neural networks
碩士 === 輔仁大學 === 資訊工程學系碩士班 === 102 === Image de-noise is the first step in image processing. Salt and pepper noises are common image noises. There are already many de-noise algorithms， such as the non local means (NL-means)， mean filtering， and median filtering. Although these filters can de-noise im...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2014
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/45991084426879623188 |
| Summary: | 碩士 === 輔仁大學 === 資訊工程學系碩士班 === 102 === Image de-noise is the first step in image processing. Salt and pepper noises are common image noises. There are already many de-noise algorithms， such as the non local means (NL-means)， mean filtering， and median filtering. Although these filters can de-noise images， they may reduce image details at the same time， resulting image blur and distortion. To reduce image blur and distortion after de-noising， we use pulse-coupled neural network PCNN (Pulse Coupled Neural Network) to de-noise images. PCNN can effectively remove noises and preserve image details.
PCNN generally uses a fixed pane size in image de-noise. It uses gray-scale values of pixels as input neurons to calculate whether pixels have noises. In this paper， dynamic sized panes are used instead of fixed sized panes. When there are no noises in a pane， our improved PCNN will automatically enlarge the size of the pane， to recalculate whether there are noises.
Keywords: Image De-noising， PCNN (Pulse Coupled Neural Network)， detect noise
|
|---|