Virtual Fitting Pose Alignment via 3D Model Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual trial fitting systems face challenges in accurately superimposing clothing images onto subjects due to differences in body shape and pose between the subject and the dummy used to create the clothing images, leading to inaccuracies in composite images displayed during virtual fitting.
Innovation Solution
An image processing apparatus that acquires and compares three-dimensional models of the subject and target poses, calculates evaluation values for pose differences, and generates notification information to guide the subject into matching the target pose, ensuring accurate alignment for high-quality composite images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If clothing images are pre-produced by scanning a dummy wearing clothing, then the clothing images can be prepared in advance for virtual trial fitting, but the pose of the subject during virtual trial fitting will not coincide with the pose of the dummy when clothing images were produced, leading to inaccurate composite images
Solution Approach 1:
The system changes the pose parameter of the subject by providing visual feedback and guidance to help the subject adjust their pose to match the target pose corresponding to the pre-produced clothing images, thereby resolving the pose mismatch problem while maintaining the ease of pre-producing clothing images
Solution Approach 2:
The system calculates an evaluation value indicating the pose difference between the subject and the dummy, and provides notification to the subject to adjust their pose. This feedback mechanism enables the subject to iteratively adjust their pose until it coincides with the target pose, ensuring accurate composite images while maintaining the pre-production advantage
2Manufacturing precision
If the pose of the subject is guided to coincide with the predetermined pose corresponding to prepared clothing images, then the accuracy of composite images is improved, but the complexity of the system increases due to the additional guidance mechanism
Solution Approach 1:
The system replaces complex mechanical pose adjustment mechanisms with an information-based approach, using visual feedback and evaluation values to guide the subject's pose adjustment, thereby achieving accurate pose alignment without adding mechanical complexity
Solution Approach 2:
The subject themselves performs the pose adjustment based on the evaluation feedback provided by the system, rather than requiring an operator or complex automated mechanism to physically adjust the subject's pose, reducing system complexity while maintaining accuracy
3Adaptability or versatility
If a large number of clothing images corresponding to all poses of the dummy are generated, then the pose coverage is improved, but the quantity of data and storage requirements increase significantly
Solution Approach 1:
Instead of generating clothing images for all possible poses, the system identifies and uses only the specific target pose that corresponds to the pre-produced clothing images, reducing the number of required images while maintaining adaptability through real-time pose evaluation and guidance
Solution Approach 2:
The system extracts and focuses on the specific target pose information from the set of all possible poses, using only the essential pose data needed for accurate composite image generation, thereby reducing data quantity while maintaining versatility
Data Source
AI summary
According to one embodiment, an image processing apparatus includes a first acquisition module, a receiver, a second acquisition module, a calculator, a generator and an output module. The first acquisition module acquires first model data indicating a shape corresponding to a subject. The receiver receives pose data indicating a pose. The second acquisition module acquires second model data indicating a shape corresponding to a body shape of the subject and the pose indicated by the pose data. The calculator calculates an evaluation value based on the first and second model data. The generator generates notification information based on the evaluation value. The output module outputs the notification information.