Video Frame Type Detection with Packet Loss Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for frame type detection in video streams are incomplete and ineffective, especially in scenarios with packet loss, and current frame size detection is only applicable without packet loss, failing to accurately assess video quality in network environments.
Innovation Solution
A method and apparatus for frame type detection in video streams that estimate the size of a current video frame by accumulating received RTP packet payloads, estimating lost packet payloads based on adjacent packets, and determining frame type using change feature values and I-frame thresholds, even in the presence of packet loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing frame size detection method is used, then detection is simple, but it is only applicable to frames without packet loss and cannot accurately assess video quality in network environments
Solution Approach 1:
The patent performs preliminary actions by estimating the sizes of lost RTP packets before final frame size determination. It calculates the average payload size of received packets and uses this to estimate lost packet sizes, then subtracts these estimates from the total expected frame size to compensate for packet loss effects
Solution Approach 2:
The patent introduces an intermediary approach by using the average payload size of received packets as a mediator to estimate the size of lost packets. This intermediary value bridges the gap between known received data and unknown lost data, enabling frame size detection despite packet loss
2Measurement precision
If frame type detection is performed using existing methods, then the process is straightforward, but the detection is incomplete and ineffective especially in packet loss scenarios
Solution Approach 1:
The patent performs preliminary frame size detection and comparison with threshold values before making final frame type determination. It calculates the absolute difference between current and previous frame sizes, compares this difference against thresholds, and uses this preliminary analysis to guide the final I-frame or P-frame classification
Solution Approach 2:
The patent employs feedback mechanisms by continuously comparing current frame size characteristics with previous frame data and threshold values. The detection process uses feedback from frame size differences and packet loss patterns to iteratively refine frame type determination, improving accuracy in packet loss scenarios
3Measurement precision
If packet loss compensation is implemented, then frame size accuracy improves, but the detection process becomes more complex
Solution Approach 1:
The patent implements self-service by having the detection system use its own received packet data to generate the estimation criteria for lost packets. The system calculates the average payload size from its own received packets and uses this self-generated metric to compensate for lost packet sizes, making the system self-sufficient rather than requiring external compensation data
Data Source
Figure 1~2A
Figure 2B~2C
Figure 2D~2E
AI summary
Embodiments of the present invention provide frame type detection and frame size detection methods and apparatuses for a video stream. The video frame type detection method includes: obtaining a size of a current video frame, where the current video frame is located after the first video frame of the video stream; obtaining a change feature value of the current video frame according to the current video frame and a size of a previous first video frame adjacent to the current video frame; obtaining an I-frame threshold of the current video frame according to a video frame rate and a location of the current video frame; and determining whether the previous first video frame is an I frame according to the change feature value of the current video frame and a change feature value of the previous first video frame, the I-frame threshold of the current video frame, and the size of the current video frame, the size of the previous first video frame, and a size of a previous second video frame. The present invention provides a complete and effective frame type detection method, and therefore only one video frame before a previous first video frame is considered during detection of a frame type of the previous first video frame; in addition, impact of a video frame rate is fully considered.