Low-illumination color image enhancement method based on Retinex and convolutional neural network

Zhang Yonghua,Guo Xiaojie,Zhang Jiawan

2019-01-01

Abstract:The invention discloses a low-illumination color image enhancement method based on Retinex and a convolutional neural network. The low-illumination color image enhancement method comprises the steps:inputting a low-illumination color image into a decomposition network, and outputting a three-channel reflection image and a single-channel illumination image; inputting the reflection image and the illumination image into a reflection image recovery network, and carrying out denoising and color recovery processing to obtain a recovered reflection image; inputting the illumination image and the illumination adjustment parameters into an illumination image adjustment network, and outputting the adjusted illumination image; and finally, carrying out dot multiplication operation on the recoveredreflection image and the adjusted illumination image to obtain an enhanced image. Based on the Retinex theory, the low-illumination color image enhancement method utilizes the convolutional neural network to realize enhancement of the low-illumination image, constructs a loss function to perform constraint optimization on parameters of the convolutional neural network, so as to achieve the expected effects of low-illumination image brightness, contrast enhancement and image impression enhancement, remove the influence of noise and color distortion to a great extent, and allow a user to autonomously adjust and enhance the brightness.