Logo Mask Extraction from Video Streams Using Online-Weighted-Average Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods fail to effectively extract and isolate logo pixels from video streams, particularly in dynamic content where logo boundaries are obscured by moving objects, leading to challenges in applications like Frame Rate Conversion, video editing, and copyright infringement detection.
Innovation Solution
A method involving the generation of online-weighted-average frames and logo-boundary-score maps to differentiate logo pixels from non-logo pixels, using edge detection and recursive intensity calculations to create a logo mask that isolates logo pixels from video content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing methods are used to extract logo pixels from video streams, then the extraction process is simple, but the accuracy of logo pixel identification deteriorates when logo boundaries are obscured by moving objects
Solution Approach 1:
The extraction process is divided into multiple sequential stages: motion compensation frame generation, edge map generation, logo boundary score map generation, and logo mask generation. Each stage processes specific features (motion compensation, edges, boundary scores) to progressively refine logo pixel identification, allowing complex analysis to be broken down into manageable segments that improve accuracy without overwhelming system complexity
Solution Approach 2:
Motion compensation frames are generated in advance by comparing current frames with reference frames to predict logo positions before actual logo extraction occurs. This preliminary motion analysis prepares the system for more accurate boundary detection by pre-compensating for logo movement, improving subsequent extraction accuracy
2Measurement precision
If motion compensation and edge detection are used to improve logo boundary detection, then logo pixel identification accuracy improves, but processing time and computational complexity increase
Solution Approach 1:
Edge detection and boundary score calculations are applied selectively only to regions identified as potential logo boundaries through motion compensation analysis, rather than processing the entire frame. This localized approach concentrates computational resources on critical boundary regions, improving detection accuracy while reducing overall processing time
Solution Approach 2:
The system generates multiple types of maps (motion compensation frames, edge maps, boundary score maps) that provide more information than strictly necessary, but this excessive analysis is performed only on boundary regions. The additional computational effort in these localized areas significantly improves boundary detection accuracy without proportionally increasing total processing time
3Manufacturing precision
If a detailed logo boundary score map is generated using multiple frames and edge detection, then logo mask accuracy improves, but device complexity and processing requirements increase
Solution Approach 1:
The complex task of logo mask generation is segmented into distinct processing modules: motion compensation module, edge detection module, boundary score calculation module, and mask generation module. Each module handles a specific aspect of the analysis and outputs intermediate results that are passed to the next module, making the overall complex system more manageable and maintainable while achieving high accuracy
Solution Approach 2:
Boundary score maps serve as intermediary data structures that bridge edge detection results and final logo mask generation. These score maps store probabilistic information about logo boundary likelihood at each pixel location, allowing the system to make informed decisions about logo pixel classification without requiring direct complex interactions between all processing components
Data Source
AI summary
In a method of extracting a logo mask from a video stream having a logo, a video stream including a plurality of frames is received. A plurality of online-weighted-average frames is sequentially generated using the video stream. A logo-boundary-score map is generated using a current frame among the plurality of frames and a current online-weighted-average frame among the plurality of online-weighted-average frames. The logo-boundary-score map includes a plurality of logo-boundary-score entries, each of which corresponds to one of a plurality of pixels of the video stream and includes a probability that a corresponding one of the plurality of pixels is a logo-boundary pixel. The logo mask includes a plurality of entries, each of which corresponds to one of the plurality of pixels and represents whether a corresponding one of the plurality of pixels is a logo pixel or a non-logo pixel.