Objective To propose an image enhancement method based on the low-level visual features according to their weights in the different types of image. Methods Firstly the low-level visual features of the image were extracted, such as color, intensity, orientation, texture and edge features, and then computing saliency map was produced by weighted integration. The optimal weights of the low-level visual features were then determined by comparison with eye tracker measured human saliency map using the correlation coefficient. Finally, the appropriate image enhancement methods were selected based on the weights for different types of images, in order to improve the consistency of visual perception. Results Experimental results showed that the image enhancement method based on low-level visual features used in this paper had a more consistent enhanced effect with human visual perception. Conclusion We need to fully take into account the feature of the low-level visual features for different types of images, so that it can truly reflect the visual area of interest in order to improve the consistency of visual perception after image enhancement.