Encoding, decoding, and representing high dynamic range images

A method, comprising:

• parsing an image file comprising a tone-mapped (TM) base image and HDR reconstruction data, the HDR reconstruction data comprising quantized luminance ratio values and quantized residual values in color channels of a color space;
• the TM base image comprising one or more color alterations that are not recoverable from the TM base image with a luminance ratio image;
• extracting quantization parameters relating to the quantized luminance ratio values and the quantized residual values in the color channels of the color space;
• converting, based at least in part on the quantization parameters, the quantized luminance ratio values and the quantized residual values into luminance ratio values and residual values in the color channels of the color space;
• and
• reconstructing an HDR image using the TM base image and the luminance ratio values and residual values in the color channels of the color space.

A method for performing white balance correction on initial image data, said initial image data comprising tone-mapped (TM) image data and residual image data, said residual image data further comprising high dynamic range (HDR) image data, effective linear sensor values and white balance multipliers, the steps of said method comprising:

• recovering the effective linear sensor values and the white balance multipliers from said residual image data;
• producing sensor color space values from said effective linear sensor values and said initial image data to form a first intermediate image data;
• for each color channel, applying white balance multipliers to said first intermediate image data to produce a second intermediate image data;
• clamping said second intermediate image data to minimum of the maximum color channel values if there is no HDR image data available from said initial image data to form a third intermediate image data;
• and
• if clamping was not performed, transforming said second intermediate image data into a target monitor color space to form a final image data;
• and
• if clamping was performed, transforming said third intermediate image data into a target monitor color space to form the final image data.

A method for processing initial image data, the steps of said method comprising:

• inputting said initial image data;
• inputting sensor color space data and white balance multipliers;
• if a new white point for said initial image data is assumed, computing new white balance multipliers to produce current white balance multipliers;
• for each pixel in said initial image data, transforming said initial image data into a first intermediate image data back into said sensor color space;
• for each color channel in said intermediate image data, applying the current white balance multipliers to produce a second intermediate image data;
• clamping said second intermediate image data to the minimum value amount all color channel maximum values to produce a third intermediate image data;
• and
• transforming said third intermediate image data into a desired monitor color space to produce a fourth intermediate image data.