Phase Difference Detection Using Signal Variance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional phase difference detection methods in imaging apparatuses face errors due to gain differences between image signals, particularly at positions away from the optical axis, leading to incorrect focus detection results due to vignetting and manufacturing deviations.
Innovation Solution
A phase difference detection device with a photoelectric conversion unit that generates image signals from multiple pixel positions, compares these signals, calculates fluctuations, and determines phase differences based on variance to accurately assess focus state, reducing errors caused by gain differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If phase difference detection is performed using conventional amount of correlation methods, then focus detection can be achieved, but detection errors occur due to gain differences between image signals caused by vignetting and manufacturing deviations
Solution Approach 1:
The patent changes the parameter used for phase difference detection from conventional amount of correlation to variance of image signals. By using variance as the detection parameter, the system can accurately detect phase differences even when gain differences exist between image signals, thereby resolving the contradiction between measurement precision and reliability under gain difference conditions
Solution Approach 2:
The patent introduces variance calculation as an intermediary step between image signal acquisition and phase difference detection. This intermediary process transforms the raw image signals into variance values that are immune to gain differences, enabling accurate phase difference detection without being affected by vignetting or manufacturing deviations
2Area of stationary object
If image signals are obtained from pixel positions away from the optical axis to increase detection coverage, then more area can be monitored, but gain differences increase due to vignetting
Solution Approach 1:
By changing the detection parameter from amount of correlation to variance, the patent enables the use of image signals from broader areas including those away from the optical axis. The variance parameter inherently compensates for gain differences caused by vignetting, allowing expanded detection coverage without sacrificing signal uniformity
Solution Approach 2:
The patent applies variance calculation locally to each pixel position, allowing each detection point to be processed according to its local characteristics. This local processing approach maintains detection accuracy across the entire detection area, including regions with significant vignetting effects
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution effectively reduces phase difference detection errors even with image signals having gain differences, ensuring accurate focus detection by using variance as a correlation metric and re-shading correction to improve signal coincidence.
Implementation Method 1
a photoelectric conversion unit in which a plurality of pixels for photoelectrically converting received light are arrayed, the photoelectric conversion unit being configured to photoelectrically convert at least a pair of optical images received through a lens so as to generate a first image signal and a second image signal
Data Source
AI summary
The ratio between an image A and an image B is calculated as the comparison result of the images A and B obtained from a pair of optical images. The variance is then calculated in order to evaluate the statistical fluctuation of the ratio obtained for each pixel. The fluctuation due to the variance is evaluated for each phase difference while the phase between the image A and the image B is shifted. The phase difference detection is performed on the basis of the evaluation result of the fluctuation.