Deep Learning Employ for Low-Light Image Enhancement
Journal: International Journal of Science and Research (IJSR) (Vol.8, No. 8)Publication Date: 2019-08-05
Authors : Huda Dhari; Jameelah Harbi;
Page : 994-999
Keywords : Deep Leaning; Convolutional Neural Network; Low-light image; Retinex;
Abstract
Images captured under low light situations suffer from low contrast and low visibility which can effect in bad manner on computer tasks to overcome this problem, enhancing low light image needed as pre-processing. This paper is presented a trainable model for low light image enhancing. It is based on multi scale Retinex by using deep learning and convolutional neural network (CNN) algorithm. Public (LOL) dataset has been used to train this model, consisted from 500 colored, low light images. Convolutional neural network bullied-up from eleven layers. SSIM and PSNR has been used to evaluate this model showing that average value of SSIM is (o.8) and average value of PSNR is and (21).
Other Latest Articles
- The Role of the Archaeological Artist Tariq Mazloum in Preserving the Identity of the Holy Shrines of Iraq and its Totality the Kufa Mosque is a Model
- Metastasis to Cerebellopontine Angle: A Case Report
- Role of Youth Organization and Governance in Assam with Special Reference to All Assam Student Union (AASU)
- DUST Removal Framework Based on Improved Multiple Sequence Alignment Technique
- Structural Equation Model (SEM) on Causative Factors of Deviancy in Relation with Self Esteem and Academic Achievement of Deviant Students
Last modified: 2021-06-28 18:22:28