Human visual system is capable of seeing over five orders in magnitude simultaneously and can gradually adapt to scenes with high dynamic ranges of over nine orders in magnitude. The current display device, such as CRT, can not reproduce the dynamic range more than 100: 1. This paper proposes a novel method to improve the image appearance based on Retinex model using
Integrated Surround. The first novelty is that several number of surround images are integrated and unified to a single surround. The integrated surround is applied to
Center/Surround Single-Scale Retinex (
SSR) model, which reduces a “
banding artifact” seen in normal
SSR and simplifies the complicated computational steps in conventional
Multi-Scale Retinex (
MSR). The second novelty is its
fast algorithm to generate the surround images. We introduced
Gaussian Pyramid method to cut the computation time for generating a large-scale surround by tracing a “
reduction” and “
expansion” sequences using
down-sampling and
up-sampling followed by linear interpolation. The computational expense is dramatically saved less than 1/100 for getting a surround by
Gaussian convolution with large-kernel size. The proposed model worked stable to compress the dynamic range and to improve the visibility in heavy shadow areas of natural color images.
The paper discusses the scales and weights to integrate the surround images and presents the color reproducibility measured for a synthesized target image on monitor screen visually matched to a real scene set in our experimental room.
View full abstract