Face Detection via DCT Mosaic Dimensionality Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face detection algorithms for low-capacity devices face inefficiencies due to high processing times and memory requirements, particularly when dealing with varying face sizes, angles, lighting conditions, and noise, as well as limitations in real-time processing and storage capacity.
Innovation Solution
A face detection apparatus and method that reduces processing time by converting high-dimensional edge images into low-dimensional mosaic images, using a Support Vector Machine (SVM) to generate a face detection reference value, and comparing this value with input images to efficiently detect faces, thereby minimizing vector dimensionality and storage needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If template-based face detection is used to detect faces of various sizes, then face detection coverage is improved, but processing time increases significantly due to step-by-step image resizing
Solution Approach 1:
The patent transforms the face detection problem from spatial domain to frequency domain by applying Discrete Cosine Transform (DCT) to convert image blocks into coefficient blocks. This dimensional transformation allows the system to work with transformed coefficients rather than original pixel data, significantly reducing computational complexity for multi-scale detection without requiring multiple resizing operations.
Solution Approach 2:
The patent pre-calculates and stores coefficient blocks from training images in a database before actual face detection occurs. During detection, the system directly compares input coefficient blocks against these pre-computed reference blocks, eliminating the need for real-time complex computations and multiple image resizing operations.
2Measurement precision
If SVM-based face detection with high-dimensional vectors is used, then detection accuracy is improved, but memory requirements and processing time increase
Solution Approach 1:
The patent extracts only the essential discriminative features by using DCT to transform image blocks into coefficient blocks, which capture the most important frequency components. This extraction process removes redundant information while retaining the critical features needed for accurate face detection, thereby reducing memory requirements and processing time.
3Measurement precision
If feature-based face detection is used, then detection of unchangeable features is improved, but efficiency decreases due to color and contrast changes under varying light spectra
Solution Approach 1:
The patent changes the parameter space from color values (RGB) to frequency domain coefficients (DCT). This parameter transformation makes the detection process invariant to color and contrast changes caused by varying light spectra, as DCT coefficients represent structural information rather than absolute color values, thereby maintaining both accuracy and efficiency.
Data Source
AI summary
Disclosed is an apparatus and method for detecting a face from an image. The apparatus and method uses color components and enables a Support Vector Machine (SVM) having superior recognition performance to previously learn face and non-face images and determine whether an image is a face image based on a learned image database by reducing the size of a feature vector of a face as compared to conventional systems. Accordingly, the apparatus converts a face image into a mosaic image having a minimum size to reduce the dimension of the feature vector, in order to rapidly and correctly detect a face image.