Abstract

A new tone-mapping algorithm is presented for visualization of high dynamic range (HDR) images on low dynamic range (LDR) displays. In the first step, the real-world pixel intensities of the HDR image are transformed to a perceptual domain using the perceptual-quantizer (PQ). This is followed by construction of the histogram of the luminance channel. Tone-mapping curve is generated from the cumulative histogram. It is known that histogram-based tone-mapping approaches can lead to excessive stretching of contrast in highly populated bins, whereas the pixels in sparse bins can suffer from excessive compression of contrast. We handle these issues by restricting the pixel counts in the histogram to remain below a defined limit, determined by a uniform distribution model. The proposed method is compared with state-of-the-art algorithms, using some well-known metrics that quantify the quality of tone-mapped images, and is found to have the best performance.

Highlights

  • High dynamic range (HDR) images and video have gained much popularity in recent years due to advancements in enabling technologies of capture and display

  • Compression of the dynamic range of HDR image/video is often required to visualize on screen or paper

  • We propose a new histogram-based tonemapping method for HDR images

Read more

Summary

INTRODUCTION

High dynamic range (HDR) images and video have gained much popularity in recent years due to advancements in enabling technologies of capture and display. The HDR contents that contain the real-world luminance values are transformed using the PQ curve This transformation enhances the dark regions and compresses the bright ones, in line with the working of HVS. It was mentioned earlier that histogram-based tonemapping can lead to over-compression or over-enhancement of some intensity levels, if the pixel counts of the corresponding bins are too small or too large, respectively. To resolve this issue, we impose a limit on the number of maximum pixels in a bin. A large value of k raises the truncation threshold, making the truncation less likely

EXPERIMENTAL EVALUATIONS
Findings
CONCLUSION
Full Text
Paper version not known

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 call