Abstract
It is not easy to acquire a desired high dynamic range (HDR) image directly from a camera due to the limited dynamic range of most image sensors. Therefore, generally, a post-process called HDR image reconstruction is used, which reconstructs an HDR image from a set of differently exposed images to overcome the limited dynamic range. However, conventional HDR image reconstruction methods suffer from noise factors and ghost artifacts. This is due to the fact that the input images taken with a short exposure time contain much noise in the dark regions, which contributes to increased noise in the corresponding dark regions of the reconstructed HDR image. Furthermore, since input images are acquired at different times, the images contain different motion information, which results in ghost artifacts. In this paper, we propose an HDR image reconstruction method which reduces the impact of the noise factors and prevents ghost artifacts. To reduce the influence of the noise factors, the weighting function, which determines the contribution of a certain input image to the reconstructed HDR image, is designed to adapt to the exposure time and local motions. Furthermore, the weighting function is designed to exclude ghosting regions by considering the differences of the luminance and the chrominance values between several input images. Unlike conventional methods, which generally work on a color image processed by the image processing module (IPM), the proposed method works directly on the Bayer raw image. This allows for a linear camera response function and also improves the efficiency in hardware implementation. Experimental results show that the proposed method can reconstruct high-quality Bayer patterned HDR images while being robust against ghost artifacts and noise factors.
Highlights
Image capturing devices like digital cameras and camcorders have recently improved remarkably
The performance of the proposed algorithm was tested with several BP-low dynamic range (LDR) images, which were captured with a complementary metal-oxide semiconductors (CMOS) sensor at three different exposures ( t1 = t, t2 = t/4, and t3 = t/16)
A new weighting function is proposed to be designed so that each of the Bayer patterned LDR (BP-LDR) images independently covers its corresponding region according to the radiance value
Summary
Image capturing devices like digital cameras and camcorders have recently improved remarkably. To obtain HDR images, many HDR imaging approaches utilize low dynamic range (LDR) images with different exposures [1,2,3,4,5,6,7,8,9,10,11] Most of these approaches first convert the pixel values of input images into radiance values by using the camera response function (CRF), where the CRF refers to the function that maps the radiance values of a given scene to the pixel values in the captured image, and the radiance refers to the physical quantity of light energy on each element on the sensor array. The radiance values of the input images are combined into a single HDR image using weighting functions based on the reliability of the input data
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
