Study on the HDR fusion method based on dual gain grayscale image

Abstract

For the scientific complementary metal-oxide-semiconductor detector with dual gain control function, two images can be output simultaneously at the any exposure time. The low gain image has small pixel brightness value, and most of low illumination regions are usually black pixel. However, the high illumination regions in the high gain image may be saturated and overexposed. To obtain a detailed image for different illumination, a high dynamic range image fusion method base on dual gain grayscale image was developed. First Laplacian pyramid for dual gain images was established; then fusion was performed according to the weight value of each layer and image reconstruction was carried out by the pyramid results. Finally, the high dynamic range image could be obtained through grayscale adjustment. Results showed that the high dynamic range image fusion had better performance than others in many regions, such as the buildings, the roof, the sky and cars with low illumination. The image still had high resolution under the condition of zoom. Therefore, the fusion method base on dual gain grayscale image is suitable to high and low gain image, and it provides reference for high dynamic range image generation.