Multi-resolution Motion Detection for Repetitive Surfaces
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing methods for motion detection, such as optical navigation, face difficulties with surfaces that have repetitive features, like wood grain, which interfere with accurate tracking, especially when the device is moved in certain directions.
Innovation Solution
A device that decomposes input images into multiple frequency bands, allowing for the comparison of these bands with stored reference images to determine motion values, with the ability to weight certain frequency ranges differently based on surface properties and direction of motion, effectively filtering out repetitive information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing methods are used for motion detection, then the device can operate on simple surfaces, but it fails to accurately detect motion on surfaces with repetitive features like wood grain
Solution Approach 1:
The patent segments the image processing into multiple frequency bands using wavelet transform. Instead of processing the entire image uniformly, it divides the image into different frequency components (approximation and detail coefficients), allowing selective processing of specific frequency ranges that contain useful motion information while filtering out repetitive surface patterns.
Solution Approach 2:
The patent applies different processing weights to different frequency bands and spatial regions. By assigning higher weights to frequency bands that contain motion information and lower weights to bands containing repetitive surface features, it creates a localized quality approach where each frequency band is processed according to its specific characteristics rather than treating all regions equally.
2Reliability
If conventional image comparison is used, then the processing is simple and fast, but it cannot distinguish motion from repetitive surface patterns
Solution Approach 1:
The patent performs preliminary action by transforming the image into the frequency domain before comparison. The wavelet transform is applied to both the reference and current images, decomposing them into frequency bands in advance. This preliminary transformation enables the subsequent comparison to focus on frequency-specific features, improving reliability without requiring complex real-time analysis during the comparison phase.
Solution Approach 2:
The patent introduces frequency domain representation as an intermediary between the raw images and the motion detection process. Instead of directly comparing spatial domain images, it transforms them into frequency domain coefficients, which serve as an intermediary representation that highlights motion-related features while suppressing repetitive surface patterns. This intermediary transformation layer simplifies the subsequent comparison operation.
Data Source
AI summary
Motion detection using sub-band image processing. A device has decompose logic that decomposes an input image into composite images that comprise different frequency bands of the input image. The device also has storage coupled to the decompose logic to store composite images as reference images for comparison with a later input image. The device further has comparison logic to compare the composite images with the reference images to produce preliminary motion values for the different frequency bands. The device also has logic to determine a final motion value from the preliminary motion values.