Lecture Capture Privacy Protection via Positional Subject Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for protecting the privacy of subjects in lecture capture systems are inadequate, as they may incorrectly apply mosaic processing to the lecturer's face or fail to protect the privacy of auditors, and require costly registration of face images for all lecturers.
Innovation Solution
An information processing apparatus that identifies non-processing target subjects based on positional relationships or posture information within a dynamic image, and applies predetermined image processing only to processing target subjects, thereby ensuring accurate privacy protection for subjects other than the main subject.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If mosaic processing is applied to protect privacy of subjects in lecture capture systems, then privacy protection is improved, but the lecturer's face may be incorrectly processed or registration costs increase
Solution Approach 1:
The patent segments the image processing task by identifying and separating the lecturer from other subjects based on positional relationships. The controller determines which subject is the lecturer by analyzing position information and applies different processing rules: no mosaic processing for the lecturer, but mosaic processing for other subjects. This segmentation approach eliminates the need for comprehensive face registration while maintaining accurate privacy protection.
Solution Approach 2:
The patent applies local quality by treating different subjects differently based on their positional characteristics. The image processing apparatus applies mosaic processing only to specific local regions (other subjects) while excluding the lecturer's face from processing. This selective application of processing quality resolves the contradiction by protecting privacy where needed without incorrectly processing the lecturer or requiring full registration.
2Ease of operation
If face size is used to determine main subject, then main subject identification is simplified, but lecturer may be incorrectly identified when face size is not larger than auditors
Solution Approach 1:
The patent introduces position information as an intermediary factor to determine the main subject. Instead of relying solely on face size, the system uses the lecturer's positional relationship with the camera and other subjects as a mediating criterion. The controller uses this intermediary position data to accurately identify the lecturer even when face size does not distinguish them from auditors, resolving the contradiction between simple identification and accurate identification.
Data Source
AI summary
An information processing apparatus includes: a controller that identifies a non-processing target subject, among multiple subjects in a frame included in a dynamic image, as a non-processing target that is not subjected to predetermined image processing, on a basis of positional relationships between multiple subject areas corresponding to the respective multiple subjects and a specific area in the frame; and an image processor that performs the predetermined image processing on a processing target area corresponding to a processing target subject other than the non-processing target subject, among the multiple subjects.