High Dynamic Range Image Generation of Dynamic Scenes

To solve the failure of ghost removal resulting from the inaccuracy of ghost detection and the de- tail insufficient in the generated high dynamic range image due to the single reference image-based ghost removal, a multi-exposure high dynamic range imaging method without ghost artifacts and high detail in- formation is proposed in this paper. Firstly, a partial ghost region is obtained by the order relationship be- tween the gray values of two adjacent frames, and the other ghost region is acquired by comparing the gray values of the difference image between two adjacent images with a threshold that is optimized by multiple iterations. After separating the ghost region of the adjacent images, the new ghost region between the se- lected single reference image and the rest of the input images is integrated by these images ghost region. Then, the image inpainting method is used to enhance the detail information of the non-reference image in the ghost region after removal of the moving objects; finally, a high dynamic range image is generated using a multi-exposure high dynamic range imaging method for static scenes. Experimental results show that compared with the existing methods, the proposed method not only can effectively remove ghost artifacts, but also avoids the drawback that the ghost region detail information in the single reference image method is completely determined by the single frame reference image, and preserves and improves the details of the generated image in the ghost region.