Low Illumination Image Enhancement Algorithm Based on Convolutional Neural Network

Image enhancement processing is to improve image quality and obtain more image information. Due to the limitation of image acquisition equipment and the influence of external environment, the acquired images have some problems such as low contrast, color distortion and loss of detail information, which seriously affect the subsequent application of images. Therefore, how to achieve low illumination image enhancement and obtain high quality image has become an important research field. Based on convolutional neural network technology, this paper studies the image quality evaluation indexes such as peak signal-to-noise ratio and structural similarity, designs ResNet network structure and U-Net network structure, and studies the image enhancement effects of two types of convolutional neural networks. The research results have a good enhancement effect, and the enhanced image is closer to the real natural image, so as to improve people's experience of using the image. It is found that U-Net network is superior to ResNet network in the enhancement of low illumination image. The improved illuminance map enhancement network can enhance the illumination component of different brightness regions in the image.