Real-time Face Detection Using Pixel Pairs and Cascading Classifiers

Solution Overview

Problem

Existing Haar-wavelet-based real-time face detection algorithms require powerful computing devices and are not suited for real-time processing on mobile hardware, making them inefficient on contemporary mobile device hardware.

Innovation Solution

A face detection technology that processes image data into an image pyramid, using pixel pairs as features and a cascading classifier to efficiently detect faces, with effective contrast normalization and image padding to increase the field of view, allowing for real-time face detection on low-power devices.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If Haar-wavelet-based face detection algorithms are used, then detection accuracy is improved, but processing time increases and power consumption increases

Engineering Contradiction:

Improvedetection accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent extracts only the essential discriminative features needed for face detection by using pixel pairs instead of computing full Haar-wavelets. This selective extraction maintains detection accuracy while significantly reducing computational complexity and processing time on mobile devices.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The detection process is segmented into multiple stages with progressively more complex pixel pair comparisons. Early stages use simple comparisons for rapid filtering, while later stages use more complex patterns for final verification, enabling real-time processing without sacrificing accuracy.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If Haar-wavelet-based face detection algorithms are used, then detection accuracy is improved, but device complexity increases

Engineering Contradiction:

Improvedetection accuracyVSAvoidcomputing device complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential discriminative features needed for face detection by using pixel pairs instead of computing full Haar-wavelets. This selective extraction maintains detection accuracy while significantly reducing computational complexity and processing time on mobile devices.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent uses simple, computationally inexpensive pixel pair comparisons that can be rapidly evaluated many times throughout the cascade stages, replacing the need for complex, resource-intensive Haar-wavelet computations.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

3Productivity

If more processing power is allocated to face detection, then detection speed is improved, but power consumption increases

Engineering Contradiction:

Improvedetection speedVSAvoidpower consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The detection process is segmented into multiple stages with progressively more complex pixel pair comparisons. Early stages use simple comparisons for rapid filtering, while later stages use more complex patterns for final verification, enabling real-time processing without sacrificing accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The cascade classifier structure allows the system to perform partial processing for most candidate regions (quickly rejecting non-faces in early stages) and only applies full processing to a small subset of promising candidates, reducing overall computational load and power consumption.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS8995772B2Real-time face detection using pixel pairs

Publication Date: 2015.03.31 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

The subject disclosure is directed towards a face detection technology in which image data is classified as being a non-face image or a face image. Image data is processed into an image pyramid. Features, comprising pixel pairs of the image pyramid, are provided to stages of a cascading classifier to remove sub-window candidates that are classified as non-face sub-windows within each stage. The face detection technology continues with one or more subsequent stages to output a result as to whether the image contains a face.