A Novel Low Illumination Image Enhancement Algorithm
On the purpose of the better visibility and understanding for low illumination images, the proposed method improved the homogeneity measurement with fuzzy entropy and fractal dimension, in order to select the reasonable pixels in the pending sub-region an
- PDF / 255,520 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 40 Downloads / 193 Views
A Novel Low Illumination Image Enhancement Algorithm Lijuan Wang and Jiansheng Qian
Abstract On the purpose of the better visibility and understanding for low illumination images, the proposed method improved the homogeneity measurement with fuzzy entropy and fractal dimension, in order to select the reasonable pixels in the pending sub-region and reduce noises as possibly. It was experimented on large of low illumination images and evaluated by using the metrics for the enhanced images, the results proved that the proposed method was efficient for avoiding over-enhancement and preserving better light and dark contrast and details. Keywords Low illumination • Homogeneity entropy • Membership • Image enhancement
247.1
measurement
•
Fuzzy
Introduction
In most industrial sites with confined space, the environmental conditions is complex and poor, which is characteristic of low illumination, more dust, poorly ventilated and high dangerous, so the video monitoring system should be installed in key areas of confined space in order to ensure safety. But due to above the characteristics, the videos and images acquired have the features of uneven illumination, big mixed noise, unclear details and bad contrast, all of which shows poor visibility and reading so that it is difficult to catch the abnormal situation in the images and make correct judgment and decision for the inspectors. Thus aided by image enhancement and analysis technique is essential for detecting the abnormal situation in the videos and images.
L. Wang (*) • J. Qian School of Information and Electric Engineering, China University of Mining and Technology, Jiangsu Xuzhou 221116, China e-mail: [email protected] S. Zhong (ed.), Proceedings of the 2012 International Conference on Cybernetics 1931 and Informatics, Lecture Notes in Electrical Engineering 163, DOI 10.1007/978-1-4614-3872-4_247, # Springer Science+Business Media New York 2014
1932
L. Wang and J. Qian
For image enhancement and analysis, the main processing object is the image having physical characteristics based on human visual system, the purpose of which is to get more useful information. The algorithms of image enhancement are mostly based on the features of grayscale, edge and texture. Yiu-fai Wong [1] described how image enhancement could be well done with the consideration of the edgepreserving filter. Cheng and Shi [2] employed multi-peak histogram equalization combined with local information to enhance the image. Chen, Daponte and M.D [3] obtained the fractal dimension in medical images for Two applications are found: (1) classification; (2) edge enhancement and detection. Cheng and Hu [4] employed the fuzzy entropy principle and the fuzzy set theory to enhance the contrast. Above these methods mainly take into account local information processing to achieve image enhancement and reduce over-enhancement to a certain degree, however, in some cases it is still possible to enhance noise and make over-enhancement. Cheng, Xue and Shi [5] used the homogeneity measurement method for
Data Loading...
