[Invited Paper] An Efficient Unified-Tone-Mapping Operation for HDR Images with Various Formats

Abstract

This paper addresses a unified tone mapping operation (TMO) for HDR images with fixed-point arithmetic. A TMO generates a low dynamic range (LDR) image from a high dynamic range (HDR) image by compressing its dynamic range. A unified TMO can perform tone mapping for various HDR image formats with a single common TMO. Since HDR images are generally expressed in a floating-point data format, a TMO also deals with floating-point data even though resulting LDR images have integer data. As a result, conventional TMOs require many resources such as computational and memory cost. To reduce the resources, the method which allows to replace a floating-point number with two 8-bit integer numbers was proposed. However, this method has a limitation of available input HDR image formats. The proposed unified TMO can be applied for various formats such as the RGBE and the OpenEXR by introducing an intermediate format. Moreover, the method can conduct all calculations in the TMO with fixed-point arithmetic. By using both integer data and fixed-point arithmetic, the method reduces not only the memory cost but also the computational cost. The experimental and evaluation results show the proposed method reduces the computational and memory cost, and gives almost same quality of LDR images, compared to the conventional method with floating-point arithmetic.