Video Processing Device Pull-Down Cadence Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video processing methods, such as those using motion estimation and motion compensation, often result in motion judder, halo effects, and video latency, particularly when converting frame rates to comply with broadcasting standards like NTSC or ATSC, which require 60 FPS, and fail to accurately handle occlusion regions or high-frequency areas.
Innovation Solution
A processing device and method that detect and adjust the pull-down cadence format of video streams to linearly change the frame rate, allowing for integer multiples of the original frame rate, thereby bypassing or modifying the input video stream to reduce motion judder and soap opera effects, and improve video latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If motion estimation/motion compensation method is used to smooth motion by interpolating new frames, then motion judder is reduced, but side effects like halo effect occur and video latency increases
Solution Approach 1:
The patent extracts and removes repeated frames from the video sequence based on detected pull-down cadence patterns, eliminating the need for motion estimation and compensation. This approach avoids the harmful side effects (halo effect, video latency) while maintaining motion smoothness by naturally eliminating the source of motion judder through frame selection rather than frame interpolation
Solution Approach 2:
The patent uses frame copying/repetition detection to identify pull-down patterns by comparing consecutive frames. Instead of using complex ME/MC algorithms, it copies and compares frame data to detect cadence patterns, then uses this information to selectively output frames, achieving motion smoothing without the harmful artifacts of traditional interpolation methods
2Speed
If pull-down method is applied to convert frame rate to comply with broadcasting standard, then frame rate conversion is achieved, but specific frames are repeated more than others causing motion judder
Solution Approach 1:
The patent dynamically adjusts frame output based on detected pull-down cadence patterns. Instead of applying a fixed pull-down method that uniformly repeats frames, it dynamically identifies which frames are repeated based on motion analysis and cadence detection, then selectively outputs frames to maintain consistent timing. This dynamic approach eliminates motion judder while achieving the required frame rate conversion
Solution Approach 2:
The patent changes the frame selection parameters based on detected pull-down patterns. By analyzing motion vectors and frame differences, it identifies the specific cadence pattern (e.g., 3:2, 2:3:3:2) and adjusts which frames to output accordingly. This parameter-based frame selection eliminates uneven frame repetition and motion judder while maintaining the target frame rate
3Productivity
If motion estimation/motion compensation is used to interpolate frames, then frame rate is increased, but significant video latency occurs
Solution Approach 1:
The patent extracts and removes redundant repeated frames rather than adding interpolated frames through ME/MC. By detecting pull-down cadence patterns and eliminating the extra frames that cause judder, it achieves smooth motion playback at the original frame rate without introducing the significant video latency that comes with complex motion estimation and compensation algorithms
Solution Approach 2:
The patent skips the computationally intensive motion estimation and compensation steps entirely. Instead of performing ME/MC to interpolate frames, it directly detects pull-down patterns and skips outputting the identified repeated frames, achieving the same motion smoothing effect with minimal processing delay and significantly reduced video latency
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided are a processing device and a method of outputting a video. The processing device includes a pull-down cadence detector (410) configured to receive an input video stream (IVS) generated by pulling down raw data from an external device, detect a pull-down cadence format indicating a rate at which the raw data is pulled down based on the input video stream, and output detection result information (DRI); and a pull-down cadence controller (420) configured to output an output video stream (OVS) having a second frame rate that is an integer multiple of a first frame rate of the raw data, based on the input video stream (IVS) and the detection result information (DRI).