DASH Video Playback Access Accuracy via Segment Index SAP Location
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing Dynamic Adaptive Streaming over HTTP (DASH) technology experiences poor access accuracy when transmitting video streams based on background modeling coding, due to the long GOP structure, which results in a large difference between the user-input access time point and the actual access point, affecting user experience.
Innovation Solution
The solution involves adding location information of independently decoded frames and non-independently decoded frames to the segment index, allowing the terminal to determine the video frame closest to the access time point, and starting decoding from that frame for improved access accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the length of GOP is increased to improve compression rate, then coding efficiency is improved, but error diffusion effect deteriorates
Solution Approach 1:
The patent segments the GOP structure into multiple independent SAP points within a single GOP, allowing the video stream to be divided into multiple accessible segments. Each SAP point represents an independent decoding start point, enabling the system to maintain long GOP lengths for compression while providing multiple entry points that limit error propagation scope.
2Device complexity
If only one SAP is defined for the current GOP, then the structure is simple, but access accuracy is poor
Solution Approach 1:
The patent dynamically determines the optimal SAP point based on the user's access time point. Instead of using a fixed single SAP, the system identifies which SAP point (among multiple predefined points in the GOP) is closest to the requested access time, allowing flexible and accurate random access while maintaining a relatively simple overall structure.
3Ease of operation
If there is a large difference between access time point and SAP location, then decoding can start from any point, but user experience is affected
Solution Approach 1:
The patent performs preliminary action by pre-defining multiple SAP points within each GOP structure during encoding. These predetermined access points are strategically placed to minimize the maximum distance from any user-requested time point to the nearest SAP. This preliminary structuring ensures that when a user requests access at any time point, the system can quickly identify and start decoding from the closest pre-prepared SAP point, minimizing delay and maintaining good user experience.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The present invention provides a network video playback method and an apparatus, where the method includes: receiving an MPD file of a video, and acquiring an access time point requested by a user; determining, according to the MPD file, a media segment corresponding to the access time point; acquiring a segment index of the media segment, where the segment index includes location information of an independently decoded frame in the media segment and location information of a non-independently decoded frame that is in the media segment and that references an independently decoded frame; determining, according to the location information of an independently decoded frame in the media segment and location information of a non-independently decoded frame that is in the media segment and that references an independently decoded frame , a video frame closest to the access time point from the independently decoded frame and the non-independently decoded frame; if the video frame closest to the access time point is a non-independently decoded frame, acquiring an independently decoded frame that is referenced by the video frame closest to the access time point; and starting decoding from the video frame closest to the access time point according to the referenced independently decoded frame, and performing video playback. According to the present invention, access accuracy of network video playback can be improved.