Image Converting Module Segment-Based Tone Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional display devices have a limited dynamic range, unable to effectively display high dynamic range images, as they are typically designed for low dynamic range formats, leading to a mismatch between the display's capabilities and the human vision system's ability to distinguish a wide range of colors and luminance levels.
Innovation Solution
An image converting method and module that calculates segment averages and local averages of pixel values to generate a specular map and confidence, mapping and blending input images to create an intermediate high dynamic range image, reducing computing costs and improving power efficiency by analyzing segments rather than pixels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional display devices display LDR images, then the display device can show images within its limited dynamic range, but the display cannot effectively utilize the human vision system's ability to distinguish a wide range of colors and luminance levels
Solution Approach 1:
The patent transforms the luminance parameter distribution by mapping LDR pixel values to HDR luminance values using tone mapping functions. This parameter transformation expands the displayed luminance range from 0-255 to a much wider HDR range, enabling the display to utilize its full dynamic range capability while preserving visual information through perceptual scaling
2Measurement precision
If image converting method analyzes each pixel individually, then the conversion precision can be maintained, but the computing cost and power consumption increase significantly
Solution Approach 1:
The patent divides the image into multiple local regions or segments and performs tone mapping operations on segmented blocks rather than individual pixels. This segmentation approach maintains sufficient conversion accuracy within each local region while dramatically reducing the computational complexity and power consumption compared to processing every pixel independently across the entire image
Data Source
AI summary
An image converting method includes calculating a plurality of segment averages of pixel values in a plurality segments of an input image; acquiring the maximum among the segment averages as a first threshold; calculating a plurality of local averages of pixel values in adjacent segments of input image; acquiring the maximum among local averages as a second threshold; counting the number of pixel values exceeding first threshold in the segments as a plurality of first pixel counts and counting the number of pixel values exceeding second threshold in the segments as a plurality of second pixel counts; generating a specular map and a confidence according to the first pixel counts and the second pixel counts; mapping the input image according to the specular map, to generate an intermediate image; and blending the input image and the intermediate image according to the confidence, to generate an output image.