Abstract
As ubiquitous computing integrates technology into daily life, the demand for non-intrusive, human-friendly interfaces grows. This paper presents the Half-Mirror Interface System (HMIS), a novel man-machine interface that utilizes a half-mirror to transition between a functional mirror and a digital display. Equipped with a webcam, microphone, and speakers, the HMIS offers personalized services such as makeup assistance, weather updates, and health diagnostics based on Sasang constitutional medicine. Experimental results show high accuracy in constitutional classification (89.7%), demonstrating the system's potential as a versatile home interface.
Introduction
The paradigm of ubiquitous computing envisions technology embedded seamlessly into the human environment. However, traditional interfaces like monitors and keyboards often clash with domestic aesthetics. The HMIS addresses this by repurposing the mirror—a common household object—into a multi-functional interface. By leveraging the optical properties of a half-mirror, the system provides utility when the user is present and serves as a decorative element when idle.
System Design and Operation
The HMIS hardware comprises a half-mirror, an LCD panel, a USB webcam, a microphone, and speakers. The system's behavior is governed by two primary modes, determined by real-time face detection.
Principal Modes
- Mirror-Mode (User Present): Activated upon face detection. It offers makeup assistance (magnifying facial features with TTS guidance), weather updates, and health diagnostics. Hands-free operation is enabled via voice recognition software.
- Display-Mode (User Absent): In the absence of a user, the system transitions to display-mode. The LCD panel displays decorative art, family photos, or "healing windows" accompanied by music designed for psychological relaxation.
Health Monitoring and Constitutional Classification
A key innovation of the HMIS is its health monitoring engine based on Sasang constitutional medicine. The system uses image processing to identify facial proportions and color variances in specific zones.
A fuzzy logic system classifies the user into one of three constitutional types: Tae-eumin, So-eumin, or So-yangin. It further analyzes skin tone (e.g., checking for redness in the "heart" zone or bluish tints in the "liver" zone) to assess current health status, providing an objective quantification of traditional medical observations.
Experimental Results
The HMIS was evaluated with 100 male subjects in their 20s. For constitution classification, data from 50 subjects was used for training and 50 for testing. The system achieved an average accuracy of 89.7% (100% for Tae-eumin, 85.7% for So-eumin, and 83.3% for So-yangin). This performance significantly exceeds the ~60% accuracy typical of traditional survey-based classification methods. Additionally, the system's health diagnostic mapping was verified against expert medical knowledge using reference facial images representing various health conditions.
Discussion and Future Work
While the HMIS offers a promising "human-friendly" interface, several challenges remain. The current color-based diagnostic system is sensitive to ambient lighting, which can vary significantly in home environments. Furthermore, the initial study's demographic focus on young men necessitates future work to include broader age groups and diverse skin tones. Future iterations will also incorporate larger clinical databases to refine the diagnostic accuracy of the fuzzy logic engine.
Summary updated