• M.Saritha Devi
• K.Aswini Satya
• K.V.V.Ramesh
• E.Sneha
• D.L.Vijay Daniel,J.Shyam Ganga Prasad

-Low light images often exhibit colour distortion, fuzzy features, and poor contrast as a result of light scattering and attenuation in the water. In light of these issues, we present a low lightimage convolutional neural network (CNN) that uses structure decomposition for low lightimage improvement. In this case, theoretical analysis of the low lightimaging allows for the decomposition of the raw low lightimage into high-frequency and low-frequency components. Here, we provide a novel probabilistic approach to improving images by estimating both light and reflectance in the linear domain at the same time. Our results demonstrate that, compared to the logarithmic domain, the linear domain model is superior for representing background knowledge in order to estimate reflectance and illumination more accurately. The formulation is a maximum a posteriori (MAP) using previous knowledge about the lighting and the reflection. An asymmetrical multiplier approach is used to compute illuminance and reflectance for the MAP issue. The experimental findings demonstrate the potential convergence rate of the suggested approach, as well as its good performance in obtaining reflectance and illumination. The suggested approach produces findings that are on par with or even superior to those obtained using other techniques of testing. The results of the ablation research prove that each component works as intended, and the results of the application tests show that the various approaches can produce high-quality photos in low light.

LIME,CNN,Light image,HE,MAP

"IMAGE ENHANCEMENT IN LOW LIGHT CONDITIONS EMPLOYING ILLUMINATION MAP ESTIMATION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 3, page no.b333-b342, March-2023, Available :http://www.jetir.org/papers/JETIR2303141.pdf

"IMAGE ENHANCEMENT IN LOW LIGHT CONDITIONS EMPLOYING ILLUMINATION MAP ESTIMATION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 3, page no. ppb333-b342, March-2023, Available at : http://www.jetir.org/papers/JETIR2303141.pdf