Gradient Phase Map Feature Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional feature detection in image sensors requires significant memory and processing resources, leading to time delays and increased power consumption due to the need for large numbers of classifiers to identify features like smiles and blinks.
Innovation Solution
The method involves generating a gradient phase map of image pixel intensities and applying a projection function to determine the state of features, such as eyes or mouths, using processing circuitry within the camera, which reduces the reliance on extensive training images and computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional feature detection using large numbers of classifiers from training images is used, then feature detection accuracy is improved, but memory resources and processing resources are significantly increased
Solution Approach 1:
The patent extracts only the essential gradient phase information from images, representing features through compact gradient phase maps rather than storing complete training images or large classifier sets. This extraction of key characteristics reduces memory requirements while maintaining detection capability.
Solution Approach 2:
Instead of comparing current images against stored training images or classifiers, the patent inverts the approach by transforming both reference and current images into gradient phase space, where feature detection becomes a simpler comparison of gradient characteristics rather than full image matching.
2Reliability
If conventional feature detection using large numbers of classifiers from training images is used, then feature detection accuracy is improved, but processing resources and time delays are significantly increased
Solution Approach 1:
The patent replaces the mechanical process of comparing images against large classifier databases with a mathematical transformation approach using gradient phase analysis. This substitution of computational mechanics reduces processing complexity from O(n) image comparisons to O(1) gradient phase evaluations.
Solution Approach 2:
The patent changes the parameter space from raw pixel values and full image data to gradient phase characteristics. By transforming the detection problem into gradient phase domain, the system achieves faster computation while maintaining accuracy, as gradient phases capture essential feature information in a compressed form.
3Reliability
If conventional feature detection using large numbers of classifiers from training images is used, then feature detection accuracy is improved, but power consumption is significantly increased
Solution Approach 1:
The patent extracts only the necessary gradient phase information needed for feature detection, eliminating the need to process and compare against extensive training image datasets. This extraction approach reduces computational workload and consequently lowers power consumption while maintaining detection accuracy.
4Adaptability or versatility
If conventional feature detection using large numbers of classifiers from training images is used, then comprehensive feature identification is improved, but device complexity is significantly increased
Solution Approach 1:
The patent creates a universal gradient phase representation that can detect multiple feature types (smiles, blinks, eyes, etc.) using the same fundamental approach. Instead of requiring separate classifier systems for each feature, the gradient phase method provides a unified framework that handles diverse features through single parameter comparisons.
Data Source
AI summary
A feature detection process includes identifying an approximate location of a feature in a preliminary image. A gradient phase map of image pixel intensities within the approximate location is computed. A projection result is determined by applying a projection function to the gradient phase map. The projection result is analyzed to determine a state of the feature.