Mid-shot Face Detection Using Geometric Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face detection methods for digital images, such as the Viola-Jones algorithm, are computationally expensive and inefficient for mid-shot images, requiring complex calculations that consume significant resources.
Innovation Solution
A method that uses simple geometric analysis to locate faces in mid-shot digital images, reducing the need for complex calculations by identifying flesh tones in the foreground, which is less computationally intensive and more efficient.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Viola-Jones algorithm with cascade classifiers and integral image is used for face detection, then face detection accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent segments the face detection task into two distinct phases: a coarse detection phase using simple geometric analysis to identify potential face regions, and a fine detection phase using flesh tone analysis to confirm and refine face locations. This segmentation allows the system to avoid applying computationally expensive algorithms to the entire image, thereby reducing overall computational complexity while maintaining detection accuracy.
Solution Approach 2:
The patent applies partial action by using simple geometric analysis as a preliminary step to identify only the regions that likely contain faces, rather than applying complex classifiers to the entire image. This partial application of simplified methods before selective use of more rigorous analysis reduces computational burden while preserving detection accuracy in the regions that matter most.
2Reliability
If Viola-Jones algorithm is applied to entire image with incremental sub-window shifting, then comprehensive face detection is achieved, but processing speed decreases
Solution Approach 1:
The patent performs preliminary geometric analysis of the image to identify regions that likely contain faces before applying more rigorous detection methods. This preliminary action creates a restricted search space that maintains detection completeness while dramatically reducing the number of computations required, thereby improving processing speed without sacrificing reliability.
Solution Approach 2:
The detection process is segmented into multiple stages: first using simple geometric properties to identify potential face regions, then applying flesh tone analysis only to those identified regions. This multi-stage segmentation ensures comprehensive face detection while avoiding unnecessary computations in regions that do not contain faces, thus improving processing speed.
3Measurement precision
If complex cascade classifiers are used for face detection, then detection accuracy is improved, but energy consumption increases
Solution Approach 1:
The patent applies partial action by using computationally simple geometric analysis for the majority of the image, and reserves energy-intensive flesh tone analysis only for regions identified as potential face locations. This partial application strategy maintains detection accuracy while significantly reducing overall energy consumption compared to applying complex classifiers uniformly across the entire image.
Solution Approach 2:
The patent performs preliminary geometric analysis that requires minimal energy to identify potential face regions, and only then applies energy-consuming flesh tone analysis to those specific regions. This preliminary filtering action reduces the total energy required for face detection by avoiding unnecessary computations in non-face areas.
Data Source
AI summary
A method for detecting a face in a mid-shot digital image of a person comprises capturing first and second mid-shot digital images of nominally the same scene using different capture settings such that the foreground is differently differentiated from the background in each image, and comparing the first and second images to determine the foreground region of the images. A portion of the foreground region likely to correspond to a face is estimated based upon the geometry of the foreground region.