Auto-focus Sensor Segmentation for Phase Detection in Low Light
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current auto-focus systems in digital cameras face challenges in accurately detecting phase differences and estimating motion due to out-of-focus blur and reduced light reception by pixels covered with black masks, leading to difficulties in achieving a reliable in-focus state.
Innovation Solution
An auto-focus system and method that utilizes a sensing unit and auto-focus control unit to extract a phase difference image, extract strong features resistant to noise and out-of-focus blur, and estimate motion vectors for lens shift using phase correlation, block matching, and hierarchical interpolation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pixels are covered with black masks to receive different phases for phase difference measurement, then phase difference detection is enabled, but the amount of light received by measuring point pixels is reduced
Solution Approach 1:
The pixel array is segmented into measuring point pixels (with black masks for phase detection) and non-measuring point pixels (without masks for normal imaging). This segmentation allows different functional zones within the same sensor array, enabling phase difference measurement while preserving light reception capability in other areas.
Solution Approach 2:
Different regions of the pixel array have different properties: measuring point pixels have black masks to create phase differences, while non-measuring point pixels have no masks to maximize light reception. This local differentiation resolves the contradiction by applying the black mask property only where phase detection is needed.
2Measurement precision
If black masks are mounted on pixels to define different phases, then phase difference measurement is possible, but out-of-focus blur exists in the input image
Solution Approach 1:
The system segments the image processing into two streams: phase difference calculation using measuring point pixels and normal image formation using non-measuring point pixels. This allows phase measurement functionality without degrading the overall image quality for the final output.
Solution Approach 2:
The patent introduces an image processing unit that acts as an intermediary, selecting and combining information from both measuring point pixels and non-measuring point pixels. This mediator resolves the conflict by using phase data from masked pixels while preserving image quality from unmasked pixels.
3Measurement precision
If conventional phase difference methods are used with black masks, then phase difference can be measured, but it is difficult to detect features and acquire reliable phase difference
Solution Approach 1:
The patent merges the advantages of both measuring point pixels (with phase information) and non-measuring point pixels (with clear image information) in the image processing unit. By combining these complementary data sources, the system achieves reliable phase difference measurement while maintaining feature detectability.
Solution Approach 2:
The patent creates a composite pixel array structure combining different pixel types (measuring point and non-measuring point) with different properties. This composite structure allows the system to leverage both phase difference capability and clear image formation capability simultaneously.
Data Source
AI summary
The present invention relates to an auto-focus system and method based on a sensor, using hierarchical extraction and motion estimation for an image. The auto-focus method may include the steps of: (a) extracting a phase difference image from a sensed image; (b) extracting a robust feature image which satisfies a preset reference with respect to noise and out-of-focus blur, from the phase difference image; and (C) estimating a motion vector for lens shift from the feature image.