On-Screen Control Image Selection for Make-Up Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional make-up simulation systems face challenges in accurate detection of characteristic coordinates and latency issues when users are not directly facing the camera, leading to low-quality make-up simulation results due to the need for on-screen control and real-time processing constraints.
Innovation Solution
The system captures video frames and selects a suitable still image for make-up control by detecting characteristic coordinates, ensuring the user's face is directly facing the camera, reducing latency by using a still image for control mode operations, allowing for precise cosmetic application and improved user experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If real-time video processing is performed for make-up simulation, then the user can see make-up effects in real-time, but latency occurs due to computation requirements
Solution Approach 1:
The system pre-processes and stores multiple candidate still images with pre-detected characteristic coordinates before the user needs to perform make-up control. When the user enters control mode, the system immediately selects a suitable pre-processed image without performing detection at that moment, thereby eliminating detection latency while maintaining real-time visual feedback capability.
2Ease of operation
If the user operates the control panel while facing the monitor, then make-up control can be performed, but the user's face does not directly face the camera resulting in lower detection accuracy
Solution Approach 1:
The system captures multiple still images during a preparation phase when the user is facing the camera, pre-detects characteristic coordinates for each image, and stores them as candidates. This allows the system to have high-accuracy pre-detected coordinates ready before the user switches to control mode, resolving the conflict between operation convenience and detection accuracy.
Solution Approach 2:
The system creates a copy of the user's face image captured when facing the camera, and performs make-up simulation on this copied image during control operations. This allows the user to control make-up while facing the monitor, using the pre-captured high-quality face image as the base, thereby maintaining both detection accuracy and operational convenience.
3Reliability
If characteristic coordinates are detected from video frames during control mode, then the latest face position can be used, but detection accuracy decreases when the user is not facing the camera
Solution Approach 1:
The system performs characteristic coordinate detection in advance during a face-detection phase, storing multiple candidate images with their detected coordinates. When control mode is activated, the system selects from these pre-detected candidates rather than performing new detection, ensuring both high accuracy and smooth control operation without real-time detection delays.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Method possesses: operation mode checking step (S102) of selecting one of update mode and hold mode; when in update mode, processing-target image updating step (S108) of updating video as update processing-target image; when in hold mode, characteristic coordinate position detecting step (S104) of detecting characteristic coordinates in input video; characteristic coordinate position examining step (S105) of examining positions of the characteristic coordinates; processing-target image selecting step (S106) of selecting processing-target image; and processing-target image holding step (S109) of holding current video frame as held processing-target image.