Off-Screen Reference Image Generation for Motion Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing MPEG-2 encoding method is limited by restricted motion vectors within the screen, leading to inaccurate motion compensation when actual motion occurs outside the screen, while MPEG-4 and H.264 allow unrestricted motion vectors, but the generation of extended reference images does not consider the orientation of the reference image, resulting in shape discrepancies and increased encoding inefficiency.
Innovation Solution
The proposed solution generates an off-screen extended reference image based on the positional relationship of similar on-screen reference images, ensuring accurate representation by considering the orientation and positional similarity, thereby improving motion compensation accuracy and reducing the difference signal and motion vector code amount.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If unrestricted motion vectors are adopted to allow motion compensation from off-screen peripheral areas, then motion compensation accuracy is improved, but the accuracy of extended reference images deteriorates due to shape discrepancies
Solution Approach 1:
The patent applies local quality by generating extended reference images with different generation methods based on their location. Off-screen peripheral areas use a generation method that considers orientation and positional relationships, while on-screen areas use conventional methods. This localized approach ensures that each region has the appropriate image quality characteristics for its specific use case.
Solution Approach 2:
The patent performs preliminary action by pre-generating extended reference images for off-screen peripheral areas before motion compensation is needed. These extended reference images are generated by considering the orientation and positional relationships with similar on-screen reference images, so that when unrestricted motion vectors point to off-screen areas, accurate reference pixels are already available.
2Device complexity
If extended reference images are generated without considering orientation, then device complexity is reduced, but encoding efficiency deteriorates due to increased difference signal
Solution Approach 1:
The patent applies parameter changes by modifying the generation parameters of extended reference images based on orientation. When the orientation of similar reference images indicates a specific direction, the extension process adjusts parameters such as the sampling direction and interpolation weights to match that orientation. This results in more accurate extended reference images that reduce the difference signal and improve encoding efficiency.
3Adaptability or versatility
If motion compensation from off-screen areas is enabled, then adaptability is improved, but information amount increases due to larger motion vector code amount
Solution Approach 1:
The patent applies copying by generating off-screen peripheral area reference images through copying and interpolating from similar on-screen reference images. Instead of treating off-screen areas as completely unknown, the system copies pixel data and gradient information from similar on-screen blocks, preserving the statistical properties and reducing the information needed to encode motion vectors that reference off-screen areas.
Data Source
AI summary
A compressed dynamic image encoding device is provided, in which a motion vector is generated by searching a reference image for an image area most similar to an image area of a video input signal; a motion-compensated reference image is generated from the motion vector and the reference image; a prediction residual is generated, by subtracting the motion-compensated reference image from the video input signal; the reference image is generated, by adding the motion-compensated reference image and the result of processing performed to the prediction residual; and an encoded video output signal is generated by the processing performed to the prediction residual. The reference image comprises on-screen reference images, located inside a video display screen, and an off-screen reference image located outside the video display screen, and the off-screen reference image is generated based on the positional relationship of plural similar reference images of the on-screen reference images.