Local Feature Calculation via Rotated Vote Cells
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing local feature amount calculation methods, such as SIFT and SURF, are slow due to redundant differential calculations and extensive floating-point operations, particularly when operated on low-specification hardware like mobile terminals.
Innovation Solution
A local feature amount calculating device that extracts feature points and near-field regions, calculates direction strengths, detects main axis directions, and generates rotated vote cell patterns to eliminate the need for linear interpolation and redundant calculations, enabling high-speed local feature amount computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional local feature amount calculation methods (SIFT, SURF) are used, then rotation invariance and scale invariance are achieved, but calculation speed is slow due to redundant differential calculations and extensive floating-point operations
Solution Approach 1:
The patent extracts only the essential directional information from pixels in the near-field region and directly accumulates these directions into histogram bins, eliminating the need for complex differential calculations and floating-point operations while preserving rotation invariance through the histogram-based representation
Solution Approach 2:
The patent replaces expensive floating-point differential calculations with simple integer-based direction encoding and histogram accumulation, using computationally inexpensive operations that can be executed rapidly on low-specification hardware while achieving the same rotational robustness
2Reliability
If conventional local feature amount calculation methods (SIFT, SURF) are used, then rotation invariance is achieved, but device complexity increases due to extensive floating-point operations
Solution Approach 1:
The patent substitutes the mechanical floating-point calculation system with a simpler integer-based directional encoding and histogram accumulation system, replacing complex arithmetic operations with discrete direction binning that is inherently more suitable for hardware implementation on mobile devices
3Reliability
If rotation correction of texture pattern is performed, then rotation invariance is achieved, but calculation time increases due to linear interpolation in units of sub pixels
Solution Approach 1:
Instead of rotating the texture pattern to achieve rotation invariance, the patent inverts the approach by directly encoding pixel directions relative to the main axis and accumulating them in a rotationally invariant histogram representation, eliminating the need for actual pattern rotation and interpolation operations
Data Source
AI summary
A local feature amount calculating device includes a local feature amount calculating unit that acquires a local feature amount of a feature point for each pixel by giving a vote of a strength of the direction of a pixel, which is calculated by a direction strength calculating unit, to a vote cell to which a pixel of interest belongs out of vote cells of a rotated vote cell pattern that is acquired by rotating a reference vote cell pattern by an angle corresponding to a main axis direction along a corrected edge gradient direction that is acquired by correcting an edge gradient direction of the pixel that is calculated by a direction strength calculating unit in accordance with the main axis direction.