Privacy-Preserving Image Sensor with Local Computer Vision Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image sensors that detect objects fail to adequately protect the privacy of imaged subjects, as they can reconstruct the appearance of objects from limited image data, which is not sufficient for privacy protection.
Innovation Solution
An image sensor system with a signal processing unit that operates in either a test mode for calibration or a private computer vision mode, where it analyzes scenes using computer vision techniques to detect objects without transmitting image features, ensuring privacy by only providing signals related to detected objects to the host device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If event sensors create pixel responses only for significant changes to protect privacy, then privacy protection is improved, but the ability to detect objects accurately deteriorates
Solution Approach 1:
The patent introduces an intermediary processing system between the event sensor and the host device. The signal processing unit performs computer vision analysis locally to generate object detection signals, while a test initiation register and test mode mechanism act as intermediaries to enable calibration without compromising privacy. This intermediary layer allows the system to maintain both privacy protection and detection accuracy by processing data locally rather than transmitting raw image data.
Solution Approach 2:
The patent extracts only the necessary object detection information from the captured scene while discarding unnecessary visual details. By using event sensors to detect only significant changes and then further processing this limited data through computer vision techniques, the system extracts minimal sufficient information for object detection without transmitting complete image data, thus maintaining privacy while enabling accurate detection.
2Object-affected harmful factors
If the image sensor processes scenes internally to protect privacy, then privacy protection is improved, but the complexity of the signal processing system deteriorates
Solution Approach 1:
The patent segments the signal processing functionality into distinct modular components: an event sensor for detecting significant changes, a signal processing unit for computer vision analysis, a test initiation register for mode control, and interfaces for data transmission. This segmentation allows each component to perform its specific function independently, reducing overall system complexity while maintaining privacy protection through localized processing.
Solution Approach 2:
The patent implements dynamic mode switching through the test initiation register, allowing the system to transition between test mode and private computer vision mode based on operational requirements. This dynamic adaptability enables the system to simplify processing during calibration (test mode) while maintaining full privacy protection during normal operation (private mode), thus managing complexity based on actual usage needs.
3Object-affected harmful factors
If the system transmits only object-related signals to the host device, then privacy protection is improved, but the amount of transmitted information deteriorates
Solution Approach 1:
The patent applies partial action by transmitting only the necessary object detection signals rather than complete scene data. The signal processing unit performs computer vision analysis to identify objects and generates minimal sufficient signals containing only essential information about detected objects. This partial transmission approach maintains privacy protection while providing the host device with sufficient information for intended applications, avoiding unnecessary information loss.
Data Source
AI summary
In some embodiments, an image sensor is provided. A signal processing unit of the image sensor is configured with executable instructions that cause the signal processing unit to perform actions comprising: reading a captured scene from a pixel array; reading a value from a test initiation register; in response to determining that the value indicates a test mode: processing the captured scene to detect a region of interest associated with the value; and providing the region of interest to a first interface for transmission to a host device; and in response to determining that the value does not indicate the test mode: analyzing the captured scene using a computer vision technique to selectively generate a signal based on the analysis of the captured scene; and selectively providing the signal to a second interface for transmission to the host device.