Generation of high dynamic range images from low dynamic range images

Abstract

An approach is provided for generating a high dynamic range image from a low dynamic range image. The generation is performed using a mapping relating input data in the form of input sets of image spatial positions and a combination of color coordinates of low dynamic range pixel values associated with the image spatial positions to output data in the form of high dynamic range pixel values. The mapping is generated from a reference low dynamic range image and a corresponding reference high dynamic range image. Thus, a mapping from the low dynamic range image to a high dynamic range image is generated on the basis of corresponding reference images. The approach may be used for prediction of high dynamic range images from low dynamic range images in an encoder and decoder. A residual image may be generated and used to provide improved high dynamic range image quality.

Description

FIELD OF THE INVENTION[0001]The invention relates to generation of high dynamic range images from low dynamic range images and in particular, but not exclusively, to generation of high dynamic range video sequences from low dynamic range video sequences.BACKGROUND OF THE INVENTION[0002]Digital encoding of various source signals has become increasingly important over the last decades as digital signal representation and communication increasingly has replaced analogue representation and communication. Continuous research and development is ongoing in how to improve the quality that can be obtained from encoded images and video sequences while at the same time keeping the data rate to acceptable levels.[0003]An important factor for perceived image quality is the dynamic range that can be reproduced when an image is displayed. However, conventionally, the dynamic range of reproduced images has tended to be substantially reduced in relation to normal vision. Indeed, luminance levels enc...

AI Technical Summary

Benefits of technology

The invention is about an improved encoding system that can adjust to the characteristics of dynamic range expansion techniques performed by a decoder. This allows for a high dynamic range image to be generated from a low dynamic range image using a mapping based on reference images. The approach allows for efficient prediction and communication between the encoder and decoder without needing additional mapping data. The technical effect is an improved dynamic range image generation.

Problems solved by technology

However, conventionally, the dynamic range of reproduced images has tended to be substantially reduced in relation to normal vision.

When traditionally encoded 8-bit signals are displayed on such displays, annoying quantization and clipping artifacts may appear.

Moreover, traditional video formats offer insufficient headroom and accuracy to convey the rich information contained in new HDR imagery.

However, this would result in a high data rate.

However, prediction of HDR from LDR data tends to be difficult and relatively inaccurate.

Indeed, the relationship between corresponding LDR and HDR tends to be very complex and may often vary strongly between different parts of the image.

Consequently, prediction of HDR images from LDR images is typically very difficult and ideally requires adaptation to the specific approach used to generate the LDR image from the HDR image.

However, the approach tends to result in suboptimal results and tends to be less accurate than desired.

In particular, the use of a global reconstruction function tends to allow only a rough estimation as it cannot take into account local variations in the relationship between HDR and LDR data e.g. caused by application of a different color grading

However, although this may allow a more local prediction, the simplicity of the linear model applied often fails to accurately describe the intricate relations between LDR and HDR data, particularly in the vicinity of high-contrast and color edges.

Images