Depth Calculation Using Monochrome Extraction to Mitigate Axial Chromatic Aberration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies using the 'depth from defocus' method for calculating depth information from color images suffer from errors due to axial chromatic aberration, which affects the precision of depth measurement.
Innovation Solution
An image processing device and method that extracts signals from pixels of the same color group to generate and compare blurred images, calculating depth information based on the difference in blurring, thereby reducing the impact of axial chromatic aberration and improving precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth information is calculated using a correlation between blur sizes of color images, then depth information can be obtained from color images, but measurement precision deteriorates due to axial chromatic aberration causing color-dependent focus position differences
Solution Approach 1:
The image sensor is divided into multiple groups, each containing pixels that capture different color information (e.g., R, G, B channels). By segmenting the color image into separate color channels and processing each channel independently, the patent eliminates the harmful effect of axial chromatic aberration that affects all colors differently. This segmentation allows depth calculation to be performed on monochrome images derived from each color channel, removing the color-dependent focus position error.
Solution Approach 2:
The patent extracts monochrome images from specific color channels (e.g., green channel) of the color image. By taking out and using only the monochrome information from a single color channel rather than processing the full color image, the method removes the influence of axial chromatic aberration while preserving the depth measurement capability. This extraction approach converts the color image processing problem into a monochrome image processing problem.
2Adaptability or versatility
If multiple pixels in each group obtain different color information, then color image capture is enabled, but depth measurement precision deteriorates due to focus position differences between colors
Solution Approach 1:
The patent segments the color image data by separating it into individual color channels corresponding to different pixel groups. Each color channel is processed independently to generate monochrome images, which are then used for depth calculation. This segmentation maintains the color image capture capability while eliminating the precision loss caused by processing all colors together.
Solution Approach 2:
The patent creates monochrome image copies from the color image by extracting information from specific color channels. These monochrome copies are then used for depth measurement instead of the original color image. This copying approach allows the system to maintain full color image capture functionality while using simplified monochrome representations for accurate depth measurement.
Data Source
AI summary
An image processing device obtains a depth information image representing a distribution of depth information about an object on the basis of first and second images that have different degrees of blurring and that are captured by an image sensor including groups each including pixels that obtain different color information about the object. The image processing device includes an extracting unit configured to extract signals, each corresponding to one of the pixels of each group, from the groups and generate first and second extracted images respectively corresponding to the first and second images, and a calculating unit configured to calculate the depth information image from a difference in degree of blurring between the first and second extracted images. The signals each corresponding to one of the pixels of each group and extracted from the groups are signals corresponding to pixels that obtain the same color information.