Image Sensor Compositing With Filtering for Moving-Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques for generating high-resolution images from multiple shifted shots fail to accurately detect moving objects due to the introduction of high-frequency components and false differences in composite images.
Innovation Solution
An image processing apparatus and method that involves capturing images at non-integral multiples of the pixel pitch, generating a high-resolution image through compositing, applying a filter to reduce high-frequency components, and calculating differences between the high-resolution and original images to accurately identify moving object regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high-resolution image is generated by compositing multiple shot images, then the resolution of the image is improved, but high-frequency components are produced that cause false differences and reduce measurement precision
Solution Approach 1:
The patent segments the image processing into distinct stages: first generating a high-resolution composite image from multiple shot images, then applying a low-pass filter to remove high-frequency components, and finally calculating the difference image. This segmentation allows the system to benefit from high resolution while eliminating the harmful high-frequency artifacts that cause false detections.
Solution Approach 2:
The patent extracts and removes the harmful high-frequency components from the composite image by applying a low-pass filter. This extraction principle separates the useful high-resolution information from the harmful high-frequency artifacts, allowing the system to retain resolution benefits while eliminating false differences caused by compositing.
2Productivity
If the difference between the high-resolution image and original shot image is obtained directly, then the moving object region can be detected, but false differences arise in regions where no moving object is present
Solution Approach 1:
The patent applies a low-pass filter to the composite image before calculating the difference image. This preliminary action removes high-frequency components that would otherwise cause false differences, ensuring that subsequent difference calculation accurately reflects only actual moving objects rather than artifacts of the compositing process.
Solution Approach 2:
The low-pass filter acts as an intermediary between the composite image generation and the difference calculation. It mediates the transition by selectively removing harmful high-frequency components while preserving the useful spatial information, thereby enabling reliable moving object detection without false positives.
Data Source
AI summary
A first obtaining unit obtains a plurality of shot images shot at a plurality of shooting positions positioned such that a spacing between adjacent ones, in a first direction or a second direction orthogonal to the first direction, of the shooting positions is a non-integral multiple of a pixel pitch of an image sensor. A generating unit generates a high-resolution image having a resolution higher than a resolution of the plurality of shot images by performing compositing processing on the plurality of shot images. A filter unit applies, to the high-resolution image, a filter that reduces a high-frequency component produced by the compositing processing. A second obtaining unit performs processing for obtaining a difference between the high-resolution image to which the filter has been applied and one shot image among the plurality of shot images.