Dominating Orientation Estimation via Trigonometric Summation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating dominating orientations in digital images, such as those used in machine vision and barcode reading, rely on quantization which can lead to inaccuracies and require complex computations, limiting their speed and accuracy.
Innovation Solution
A method that calculates representative angles of gradient vectors and uses sums of sine and cosine factors to estimate dominating orientations without quantization, allowing for fast and accurate estimation suitable for hardware implementation like Field Programmable Gate Arrays (FPGAs).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If quantization is used to estimate dominating orientations, then the computation can be simplified, but the measurement precision deteriorates due to inaccuracies introduced by quantization
Solution Approach 1:
The patent changes the mathematical parameters used for orientation estimation from quantized histogram bins to continuous trigonometric functions (sine and cosine of representative angles). This allows the system to maintain computational feasibility while achieving higher precision by working in the continuous domain rather than discrete quantized bins.
2Measurement precision
If complex computations are used to improve orientation estimation accuracy, then the measurement precision improves, but the productivity decreases due to slower processing speed
Solution Approach 1:
The patent replaces complex iterative computational mechanisms with a direct trigonometric calculation approach. By using sine and cosine sums of representative angles, the system achieves high-precision orientation estimation through closed-form mathematical expressions rather than iterative or complex algorithms, thereby maintaining fast processing speed.
3Device complexity
If quantization is applied to gradient vector orientations, then the device complexity is reduced, but the measurement precision deteriorates
Solution Approach 1:
The patent transforms the parameter representation from quantized orientation bins to continuous trigonometric parameters (sine and cosine values). This parameter change allows the system to avoid quantization errors while keeping the processing complexity manageable through efficient trigonometric summation operations.
Data Source
AI summary
A method and arrangements for estimating one or more dominating orientations (αdom) in at least a part (201; 301) of a digital image (200; 300; 600). Representative angles (α1 . . . αN) representing angles of gradient vectors (g1 . . . gN) for pixels (1 . . . N) of said at least part are obtained (401). It is further obtained (402) a target number (n) of dominating orientations. It is also obtained (404) a first sum (a) comprising added sine factors based on computed sines. The sines are computed for angles that correspond to said representative angles (α1 . . . αN) multiplied with two times the target number (n). Further it is obtained (405) a second sum (b) comprising added cosine factors based on computed cosines for the same angles that said sine were computed for. Said one or more dominating orientations (αdom) are then estimated (406) based on the first sum (a), the second sum (b) and the target number (n).