Image Position Shift Detection Under Varying Exposure Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting position shift amounts between images in chronological order, such as those taken by a digital camera, face accuracy issues when exposing conditions differ between images, especially when these conditions are unknown.
Innovation Solution
An image processing apparatus and digital camera that includes a gain calculating unit, a position shift amount calculating unit, and a detecting unit, which perform geometric transformations and calculate gain values based on luminance differences to accurately detect position shifts between images, even under varying exposure conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional position shift detection methods are used, then the detection process is simple, but detection accuracy deteriorates when exposing conditions differ between images
Solution Approach 1:
The patent segments the image processing into distinct functional units: a gain calculating unit that processes luminance differences to determine exposure compensation, and a position shift amount calculating unit that computes geometric transformations. This segmentation allows each unit to specialize in specific tasks, improving overall accuracy while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent introduces an intermediary gain value calculation step that mediates between the first and second images. By calculating a gain value based on luminance differences and using this as a transformation parameter, the system creates a bridge that reconciles images with different exposing conditions before performing position shift detection, thereby improving accuracy without directly comparing the images in their original states.
2Adaptability or versatility
If exposure state matching is performed using known exposing condition information, then exposure consistency is improved, but the method cannot be applied when exposing conditions are unknown
Solution Approach 1:
The patent implements self-service by enabling the system to automatically determine exposure compensation parameters without requiring external knowledge of the actual exposing conditions. The gain calculating unit analyzes luminance differences between images and autonomously computes appropriate gain values, allowing the system to adapt to unknown exposing conditions while maintaining detection accuracy.
Solution Approach 2:
The patent dynamically changes the gain parameter based on observed luminance differences between images. Instead of using fixed or pre-known exposure parameters, the system calculates gain values that adapt to the actual conditions, transforming the detection process to accommodate varying exposing conditions while maintaining consistent and accurate position shift measurement.
3Measurement precision
If geometric transformation is applied to match exposure states, then luminance consistency is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by selecting specific pixel groups (such as skin tone regions or regions with particular luminance characteristics) for the geometric transformation rather than processing all pixels uniformly. This selective approach maintains luminance matching accuracy in critical areas while reducing overall processing time by excluding less important regions from intensive computation.
Data Source
AI summary
A proposition is to perform a detection of a position shift amount between images which are anteroposterior in a chronological order with high accuracy. Therefore, in an image processing apparatus, a unit A performs a geometric transformation on a pixel group α in an image 1 using an input position shift amount as a parameter, and calculates a gain based on a luminance difference between the pixel group α in the image 1 being performed the transformation and pixels in an image 2 existing at an identical coordinate as the pixel group α. Besides a unit B multiplies an input gain to the image 2 and calculates a position shift amount between the image 1 and the image 2. A detecting unit executes the unit A and the unit B for at least one time or more.