Dual Bitstream Video Encoding for Privacy Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face blur techniques in surveillance systems require double memory and network bandwidth for storage and streaming, and security issues arise due to the lack of adequate data protection, as private data is secured at the system level rather than the data level.
Innovation Solution
The proposed method involves encoding and decoding private data using an image processing device with a receiving circuit, masking circuit, encoding circuit, and output circuit to generate a protected video frame and output bitstream, which includes a dual bitstream approach for secure streaming and storage, where the private data layer is encrypted and only accessible with an authorized key.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If double video streams (original and protected) are transmitted, then privacy protection is achieved, but network bandwidth and memory requirements double
Solution Approach 1:
The video stream is segmented into two separate bitstreams: a first bitstream containing encoded protected video frames (with blurred private areas) and a second bitstream containing encoded original video frames. This segmentation allows the system to transmit privacy-protected video while maintaining access to original video only when needed, reducing overall bandwidth and storage requirements compared to transmitting both complete video streams simultaneously.
Solution Approach 2:
The patent extracts only the necessary protected video data into a separate first bitstream, containing only the information needed for privacy protection during normal operation. The original video data is kept separate in a second bitstream, accessible only when authorization is provided. This extraction approach eliminates the need to transmit and store redundant duplicate streams, reducing memory and bandwidth consumption.
2Ease of manufacture
If private data is secured at system level, then implementation is simple, but security is insufficient when decryption keys are leaked or videos are transferred
Solution Approach 1:
The patent moves security from the system level to the data level by implementing encryption at the bitstream level. Two separate encoded bitstreams are created with different access requirements: the first bitstream can be decoded without authorization, while the second bitstream requires an authorization key. This dimensional shift in security implementation provides robust protection even when systems are compromised, as the encryption is embedded within the data structure itself rather than relying solely on system-level controls.
3Reliability
If original video is transmitted alongside protected video, then investigation capability is maintained, but storage and streaming costs double
Solution Approach 1:
The patent implements a dynamic access model where the availability of original video data changes based on authorization. During normal operation, only the first bitstream (protected video) is transmitted and stored. When investigation is required and authorization is provided, the second bitstream (original video) becomes accessible. This dynamic approach maintains investigation capability while minimizing storage and streaming resources during normal operation, as the full original video is not continuously transmitted or stored alongside the protected version.
Data Source
AI summary
Methods and image processing devices for encoding and decoding private data are proposed. The method for encoding private data includes to receive an original video frame, mask at least one private area in the original video frame to generate a protected video frame, generate a first encoded frame by encoding the protected video frame, and generate at least one output bitstream for streaming or storage according to the first encoded frame. The method for decoding private data includes to receive at least one input video bitstream to obtain a first encoded bitstream and a second encoded bitstream, decode the first encoded bitstream to generate a protected video frame including image data associated with at least one private area, and output the protected video frame to a display queue such that the at least one private area is displayed.