HEVC ROI Encryption Using Coding Units to Limit Propagation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing HEVC/H.265 encryption methods encrypt regions wider than the actual region of interest (ROI), leading to increased complexity and reduced compression performance, and limit the domain of use due to encryption propagation affecting surrounding pixels.
Innovation Solution
Identify coding units related to the ROI using an object detection model, perform selective encryption on these units, and restrict encryption propagation by partitioning into prediction units and applying context-adaptive binary arithmetic coding (CABAC) to maintain encryption speed and compression performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If encryption is performed on a tile basis in HEVC/H.265, then encryption can be applied to regions containing ROI, but the encrypted region becomes wider than the actual ROI and surrounding areas are irregularly encrypted
Solution Approach 1:
The patent segments the tile into multiple coding units (CUs) and further divides CUs into prediction units (PUs). This hierarchical segmentation allows encryption to be applied at the PU level rather than the entire tile level, enabling precise control over the encrypted region to match the actual ROI boundaries while preserving surrounding areas.
Solution Approach 2:
The patent applies different processing to different regions by identifying ROIs and applying encryption selectively only to PUs that contain or overlap with the ROI. Surrounding areas outside the ROI are preserved without encryption, achieving local quality differentiation between encrypted and non-encrypted regions.
2Productivity
If encryption is applied to a user-designated ROI range, then encryption speed and calculation complexity are improved, but encryption propagation affects surrounding pixels
Solution Approach 1:
By segmenting the video into frames, tiles, coding units, and prediction units, the patent creates independent processing regions. Encryption propagation is restricted to within each PU boundary, preventing it from affecting surrounding pixels in other PUs. This hierarchical segmentation isolates the harmful encryption propagation effect to the minimum necessary area.
3Object-generated harmful factors
If an independent region is designated for ROI encryption, then encryption propagation is restricted, but compression performance is greatly degraded in HEVC/H.265
Solution Approach 1:
The patent dynamically adjusts the encryption scope by identifying ROIs and adapting the encryption region to match actual ROI boundaries rather than using fixed tile boundaries. This dynamic adaptation allows independent region designation at the PU level, restricting encryption propagation while maintaining compression performance by encrypting only the necessary areas.
4Stability of the object's composition
If frames are partitioned as tiles in HEVC/H.265, then the video structure is organized, but it becomes difficult to designate regions that may be independently processed
Solution Approach 1:
The patent introduces a multi-level segmentation hierarchy: frames are divided into tiles, tiles into coding units, and CUs into prediction units. This fine-grained segmentation enables independent region designation at the PU level while preserving the overall tile-based video structure, providing both structural stability and processing flexibility.
Solution Approach 2:
The patent adds a new dimension of control by introducing the PU level between the existing tile and block levels. This dimensional refinement allows independent region designation within tiles without disrupting the overall tile-based structure, enabling versatile region processing while maintaining structural organization.
Data Source
AI summary
A method and apparatus for encrypting an ROI in a HEVC/H.265 video based on a coding unit are disclosed. The method, which is performed by an encryption apparatus, may include identifying a coding unit related to an ROI that is generated for each frame as the frames that constitute a video are partitioned as a tile, and performing selective encryption on the identified coding unit.