Image Completion With Localized Hole-Filling Processes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image completion methods, particularly in virtual reality systems, are computationally costly and time-consuming due to bandwidth limitations and machine learning processes, leading to undesirable holes in images that negatively impact user immersion.
Innovation Solution
A computer-implemented method that divides image data into portions, applies different filling processes based on hole quantity and quality, using morphological operations and machine learning inference for varying hole sizes, and combines the filled portions to complete the image efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If machine learning inference techniques are used to fill holes in images, then image completion quality is improved, but computational cost and processing time increase significantly
Solution Approach 1:
The patent applies different filling processes to different image portions based on hole characteristics. Small holes are filled using a first process (e.g., simple interpolation), while large holes are filled using a second process (e.g., machine learning inference). This local differentiation ensures high quality where needed while reducing overall computational cost.
Solution Approach 2:
The image is divided into multiple portions, and holes are categorized by size and quality attributes. The patent then applies appropriate filling processes to each category separately, segmenting the complex task into manageable sub-tasks with different computational requirements.
2Measurement precision
If machine learning processes are applied to infer missing image data, then image completion accuracy is improved, but processing time increases
Solution Approach 1:
The patent determines hole quality attributes and applies different filling processes accordingly. High-quality holes (large areas) receive machine learning inference for accuracy, while low-quality holes (small areas) receive faster but less precise filling methods, optimizing the time-accuracy tradeoff locally.
Solution Approach 2:
The patent applies machine learning inference only partially to the extent necessary - specifically to large holes where it provides significant benefit. Small holes are filled using lighter methods, avoiding excessive computational expenditure on tasks where the benefit is minimal.
3Manufacturing precision
If image data is transmitted at high rates to ensure complete scene construction, then image quality is improved, but bandwidth requirements increase
Solution Approach 1:
The patent performs hole filling operations on received image data before final display. By pre-processing the image data to fill holes using appropriate methods, the system can transmit at lower rates while maintaining acceptable image quality, as the missing data is reconstructed during processing.
Data Source
AI summary
The present disclosure relates to a computer-implemented method for completing an image, the method comprising the steps of dividing data of an image to be completed into a plurality of image portions. The method entails applying a first filling process to fill a first image portion comprising a first hole, the first hole associated with a first quantity and/or a first quality; and applying a second filling process to fill a second image portion comprising a second hole, the second hole associated with a second quantity different to the first quantity and/or a second quality different to the first quality, the second process being different to first process. The method then includes combining the filled first and second image portions to complete the image.