Adaptive enhancement algorithm for low illumination images with guided filtering-Retinex based on particle swarm optimization

The low-illumination image has the defects of low brightness and weak contrast. In this paper, an improved guided filtering-Retinex adaptive enhancement algorithm is proposed for low-illumination image. Firstly, the image is converted from RGB to HSV colour space, and then the luminance component is decomposed into sub-images of each frequency band by discrete wavelet transform. Secondly, adaptive median filtering is employed to suppress noise on high-frequency sub-image. Guided filtering-Retinex algorithm is applied to improve the contrast and detail information on low-frequency sub-image. The enhanced V component is reconstructed with Hue component and Saturation component by wavelet and converted back to RGB colour space. Finally, gamma correction is adopted to increase the brightness, and the enhanced image is obtained. Since the box filter radius and regularization parameters of the guide filter have significant influences on the enhancement effect, the particle swarm optimization algorithm is utilized to determine its optimal value for the first time to ensure the enhancement effect, which can improve the brightness and contrast. Compared with the existing enhancement algorithms, the contrast and details can be improved effectively by the proposed method, the edge information is preserved while the noise is suppressed, and the distortion from Retinex is decreased. A good image visual effect is achieved.