Image Composition Apparatus Level Matching Gain
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image composition methods struggle to accurately composite images taken with different exposures, leading to unnatural images and disrupted motion detection due to uneven gain adjustments across image areas, especially when moving subjects are present.
Innovation Solution
An image composition apparatus that divides images into small regions, calculates comparison values based on brightness measurements, and sets a representative level-matching gain to adjust image levels uniformly across the entire image, accounting for exposure differences and motion detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If gain adjustment is performed separately for different level ranges, then image levels can be matched in each range, but pixel values over the boundary do not smoothly vary and unnatural images are generated
Solution Approach 1:
The patent divides the image into multiple small regions (e.g., 4x4 grid) and calculates a representative value from comparison values across these regions. This segmentation allows the system to handle different exposure levels uniformly while maintaining global consistency, resolving the contradiction between local level matching accuracy and global gradation continuity.
Solution Approach 2:
The patent merges the level matching process across different exposure ranges by using a single representative value derived from comparing small regions of low-exposure and high-exposure images. This unified approach ensures consistent gain adjustment throughout the entire image, eliminating discontinuities at boundaries while maintaining accurate level matching.
2Measurement precision
If sensitivity calibration is performed based on evaluation metering value, then correct exposure image can be acquired, but other subjects that did not move are affected by the calibration
Solution Approach 1:
The patent calculates comparison values for specific small regions containing moving subjects and uses these to determine a representative value for gain adjustment. This local quality approach allows the system to focus calibration on relevant regions while maintaining uniform application across the entire image, thereby achieving exposure accuracy without adversely affecting stationary subjects.
3Adaptability or versatility
If level matching is performed according to setting level difference, then exposure differences can be corrected, but images taken with different shutter speeds due to mechanism errors cannot be properly composited
Solution Approach 1:
The patent calculates comparison values by actually measuring the brightness levels in small regions of the input images and uses these measured differences to determine the representative value for gain adjustment. This feedback mechanism allows the system to adapt to actual exposure differences (including those caused by mechanism errors) rather than relying solely on预设 setting level differences, thereby maintaining accurate level matching under varying conditions.
Data Source
AI summary
An image composition apparatus that is capable of acquiring a high-quality composite image that keeps continuity of the gradation of the whole image. Different exposures are set for images to be taken. Each of taken images is divided into small regions. A measured value is found based on brightness values in each of the small regions. A comparison value for each of the small regions is calculated based on the measured value of a small region of one image and the measurement value of the same small region of another image. A representative value is calculated based on the comparison values calculated for the respective small regions. A level-matching gain used for compositing the image data of the images is calculated based on an exposure level difference and the representative value. The image data of the images of which the image levels are adjusted are composited by multiplying the gain.