Transim: Transfer Image Local Statistics Across EOTFS for HDR Image Applications

Despite the popularity of high dynamic range (HDR) technology in recent years, various algorithms for image and video applications are still designed and optimized for traditional standard dynamic range (SDR) data. Directly applying SDR-optimized algorithms to HDR images and video will result in significant artifacts or coding deficiency. In this work, we present a novel preprocessing method, dubbed TransIm, which transfers local statistics for the images from the desired domain (e.g. SDR) to the current domain (e.g., HDR), while maintaining its current visual presence. It is achieved by controlling the less perceivable “noise” that is orthogonal to the sparsifiable image content, using a unitary sparsifying transform. Numerical results show that the proposed TransIm can effectively transfer local patch variance from Gamma domain to Perceptual Quantizer (PQ) domain for HDR videos. We also demonstrate that the TransIm outputs are more robust to distortions and artifacts in seam carving applications.

1. Introduction

HDR video is an emerging technology that can preserve more details in both dark and bright regions, comparing to SDR [1], [2]. Great amount of HDR images and videos have been produced in recent years, which require efficient coding and compression. While new electro-optical transfer functions (EOTFs) have been standardized and widely used in industry, e.g., the PQ curve [3], [4] for HDR videos, they changed the local statistics (e.g., local patch variance) of HDR videos from those of SDR videos. There are important applications, such as content-aware retargeting [5], [6], compression [7]–[9], denoising [10], [11], watermarking [12], etc, for which the existing algorithms are usually sensitive to image local statistics. For example, the seam carving [6] results of PQ-domain HDR images usually contain distorted regions with unnatural artifacts, as shown in Fig. 1.