Dear Ghada Sandoub

CNN is mainly used for image recognition, not enhancement.

you can use other image enhancement techniques to boost the performance of CNN.

https://www.diva-portal.org/smash/get/diva2:1341096/FULLTEXT02#:~:text=In%20recent%20years%2C%20deep%20learning,are%20the%20most%20outstanding%20models.&text=In%20addition%2C%20for%20the%20data,be%20used%20for%20transfer%20learning.

Various CNN-based methods have been demonstrated to be conducive for image enhancement, showing that multi-scale retinex is equivalent to CNN with different Gaussion kernels, the MSR-net was used to learn an end-to-end mapping between a low-light and a bright image. Also, the DPED was applied to produce DLSR- from mobile-quality images by using deep residual networks trained with a composite loss function of content, color, and texture. Despite showing improvements in enhancing image quality, these method exposed limitations in processing time since the computational operations are performed in the high-resolution space.

Enhancing brightness contrast can simply be done using imadjust, CNN is not required for that

You can perform segmentation for image enhancement :: https://au.mathworks.com/matlabcentral/fileexchange/25157-image-segmentation-tutorial

Generative Adversarial Networks (GAN) based on CNN are used to get super-resolution (i.e. to improve image resolution) images. Something similar will hopefully help you.

Generative Adversarial Networks (GAN) is a machine learning method that has gained popularity, and demonstrated improved performance in the area of image enhancement (e.g: de-noising, super-resolution, etc). To get further information see the following link:

https://machinelearningmastery.com/what-is-cyclegan/