Skin Color Patch Constructs for Face Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face detection systems in portable devices face challenges in accurately and reliably detecting faces that are not presented upright in images, due to the large search space required for all possible orientations and sizes, which is impractical for real-time processing.
Innovation Solution
A data processing system that includes a skin patch identifier, a search tile generator, and a face detector, which identifies skin patches, generates search tiles based on patch orientations, and performs face detection within these tiles using binary classifiers, allowing for efficient detection of faces at various orientations and sizes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the search space is narrowed by ignoring rotations of subwindows, then real-time detection is enabled, but only substantially upright faces can be detected
Solution Approach 1:
The patent segments the search space by dividing it into multiple oriented subspaces, each dedicated to detecting faces at specific orientations. Instead of searching the entire orientation range for each face candidate, the system partitions the problem into smaller directional segments, allowing efficient real-time detection while maintaining comprehensive orientation coverage.
Solution Approach 2:
The patent introduces an orientation dimension to the traditional face detection search space. By adding this dimensional parameter and organizing search tiles according to orientation angles, the system enables detection of faces at various rotations without exponentially increasing computational complexity, thus maintaining real-time performance.
2Reliability
If the search space includes all possible combinations of location, size and rotation, then all faces can be detected, but real-time detection becomes impractical
Solution Approach 1:
The patent performs preliminary actions by pre-organizing the search space into oriented subspaces and pre-processing image data into multiple orientation representations. This preliminary structuring allows the detection algorithm to efficiently navigate the search space without examining every possible combination, thereby achieving both completeness and real-time performance.
Solution Approach 2:
The patent implements a dynamic search strategy where the system adaptively selects and focuses on relevant oriented subspaces based on image content and detected features. Rather than statically searching all orientations equally, the system dynamically adjusts the search focus to high-probability regions, improving processing efficiency while maintaining detection reliability.
Data Source
AI summary
A data processing system for performing face detection on a stream of frames of image data, the data processing system comprising: a skin patch identifier configured to identify one or more patches of skin color in a first frame and characterize each patch in the first frame using a respective patch construct of a predefined shape; a first search tile generator configured to generate one or more first search tiles from the one or more patch constructs; and a face detector configured to detect faces in the stream by performing face detection in one or more frames of the stream within the first search tiles.