Computational Imaging Depth Map via Block-Based Inverse PSF
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for determining depth maps using a single camera require multiple images, leading to long capture and processing times, high memory requirements, and are prone to errors due to limited data and movement issues.
Innovation Solution
A method that captures a multi-focus image and partitions it into blocks, determining the best inverse multi-focus point spread function (PSF) for each block to reconstruct an all-in-focus image and generate a depth map, using a single exposure with the lens moved to multiple focus positions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple images are captured at different focus positions (DFF technique), then depth map determination is achieved, but capture time and processing time increase significantly
Solution Approach 1:
The single captured image is segmented into multiple blocks, and each block is processed independently to determine depth information. This allows parallel processing of different regions, reducing overall processing time while maintaining accurate depth map determination.
Solution Approach 2:
The patent transitions from the temporal dimension (multiple images captured at different times) to the spatial dimension (multiple focus positions encoded within a single image). By capturing light rays at different focus positions simultaneously in space rather than sequentially in time, the method achieves depth information without increasing capture time.
2Measurement precision
If multiple images are captured and stored for processing, then sufficient data is available for depth determination, but memory requirements increase
Solution Approach 1:
The patent extracts only the essential depth information from the captured light rays by analyzing focus position indicators within the single image. Instead of storing multiple complete images, only the necessary data for depth calculation is retained and processed, significantly reducing memory requirements.
3Loss of time
If DFD techniques use small number of images with differing parameters, then capture time is reduced, but error susceptibility increases due to limited data
Solution Approach 1:
By dividing the image into multiple blocks and analyzing each block independently with block-specific inverse PSFs, the method extracts more depth information from the same single image. This segmentation approach increases the effective data utilization, improving reliability without requiring additional images or increasing capture time.
4Measurement precision
If multiple images are captured for depth determination, then comprehensive depth information is obtained, but device complexity increases
Solution Approach 1:
The patent combines multiple focus position measurements into a single captured image by encoding depth information through focus position indicators within the same image frame. This merging approach maintains comprehensive depth information while simplifying the system architecture, as only a single image capture mechanism is required instead of multiple sequential captures.
Data Source
AI summary
A method for all-in-focus image reconstruction and depth map generation in an imaging device is provided that includes capturing a multi-focus image by the imaging device, partitioning the multi-focus image into a plurality of blocks, determining, for each block of the plurality of blocks, a best inverse multi-focus point spread function (PSF) for reconstructing original image intensity values in the block, wherein the best inverse multi-focus PSF is selected from a plurality of predetermined inverse multi-focus PSFs stored in a memory of the imaging device, and applying to each block of the plurality of blocks the best inverse multi-focus PSF determined for the block to reconstruct the all-in-focus image.