Phase-Based Motion Detection Using Gaussian Windowed FFT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer-based motion detection algorithms, such as those using optic flow techniques, face high computational demands, making real-time implementation challenging, and there is a need for an improved technique that can efficiently detect visual motion with reduced complexity.
Innovation Solution
A phase-based motion detection method that involves constructing a multi-dimensional Gaussian window, dividing frames into blocks, multiplying each pixel with the Gaussian window, computing the multidimensional FFT to obtain amplitude and phase, and detecting motion based on the phase of the windowed signal blocks, which can be implemented using a processor and memory system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If optic flow techniques are used for motion detection, then accurate motion estimation is achieved, but computational complexity becomes too high for real-time implementation
Solution Approach 1:
The patent extracts only the essential phase information from the Fourier transform of image blocks, discarding amplitude and other non-essential data. This extraction of critical motion-related phase data reduces computational complexity while maintaining motion detection accuracy, directly resolving the contradiction between precise motion estimation and computational burden
Solution Approach 2:
The patent divides the image into multiple blocks and processes each block independently through FFT to extract local phase information. This segmentation enables parallel processing of different image regions, significantly reducing overall computational complexity while preserving accurate motion estimation through localized phase analysis
2Measurement precision
If complex motion detection algorithms are implemented, then detection accuracy is improved, but processing speed decreases
Solution Approach 1:
The patent extracts only phase information from the Fourier transform, ignoring amplitude and other complex computations. This selective extraction maintains motion detection accuracy by focusing on the most relevant data while dramatically reducing processing time and computational overhead
Solution Approach 2:
The patent transforms the motion detection problem from intensity-based analysis to phase-based analysis by computing the Fourier transform and extracting phase information. This parameter transformation enables faster processing through efficient FFT algorithms while maintaining or improving motion detection accuracy through phase congruency measures
Data Source
AI summary
Systems and methods for the detection of motion in a multi-dimensional signal are provided. According to aspects of the disclosure, data representing a time sequence of frames of a video stream is received. A multi-dimensional Gaussian window in is then constructed. Each frame of the received data is divided into a plurality of blocks and, for each block in the plurality of blocks, each pixel in the block is multiplied with a corresponding pixel of the Gaussian window to obtain a windowed video block. The FFT of the windowed video block is computed to obtain an amplitude and phase thereof. The occurrence of motion in the block is then determined based on the computed phase of the windowed video block.