Photometric Stabilization for Time-Compressed Video Flickering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Time-compressed videos often exhibit flickering due to temporal variations in color and tone caused by scene changes and automatic adjustments in imaging characteristics, which are exacerbated when played back at faster speeds, making the content visually unpleasant for viewers.
Innovation Solution
Photometric stabilization of time-compressed videos is achieved by determining correspondences between frames, computing photometric transformations, and applying filters to smooth characteristic changes, while preserving actual content changes, using a digital medium environment and frame sampling techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If conventional photometric stabilization techniques are used on time-compressed videos, then temporal variations in color and tone are reduced, but the techniques are unsuitable because they rely on significant content overlap between adjacent frames which does not exist in time-compressed videos
Solution Approach 1:
The patent changes the fundamental parameters of the photometric stabilization approach by computing transformations between non-adjacent frames rather than adjacent frames, and by using a different mathematical model (photometric transformation with occlusion handling) that adapts to the sparse sampling nature of time-compressed videos
Solution Approach 2:
The patent segments the video processing into distinct stages: frame sampling to create time-compressed video, computation of photometric transformations between selected frames, and application of filters to smooth transitions. This segmentation allows each stage to be optimized independently for time-compressed video characteristics
2Productivity
If frame sampling is used to create time-compressed video, then video duration is reduced and playback speed is increased, but temporal variations in illumination are exacerbated causing flickering
Solution Approach 1:
The patent applies photometric stabilization processing after the video has been time-compressed through frame sampling. This preliminary action of creating the time-compressed video first allows the subsequent stabilization process to specifically address the flickering artifacts introduced by the sampling process
Solution Approach 2:
The patent uses a feedback mechanism where photometric transformations are computed between frames, filters are designed based on these transformations and photometric similarity measures, and then these filters are applied to smooth the photometric variations. This closed-loop approach continuously reduces the flickering effect
Data Source
AI summary
Photometric stabilization for time-compressed video is described. Initially, video content captured by a video capturing device is time-compressed by selecting a subset of frames from the video content according to a frame sampling technique. Photometric characteristics are then stabilized across the frames of the time-compressed video. This involves determining correspondences of pixels in adjacent frames of the time-compressed video. Photometric transformations are then determined that describe how photometric characteristics (e.g., one or both of luminance and chrominance) change between the adjacent frames, given movement of objects through the captured scene. Based on the determined photometric transformations, filters are computed for smoothing photometric characteristic changes across the time-compressed video. Photometrically stabilized time-compressed video is generated from the time-compressed video by using the filters to smooth the photometric characteristic changes.