Deterministic Digital Image Matching via Parallel Wavefront Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital image editing systems face issues with reliability, flexibility, accuracy, and efficiency due to their reliance on randomized searching, leading to unpredictable and non-deterministic results, inaccuracy in propagating matching patch values, and inefficiency in processing power usage, making it difficult to generate consistent and accurate edited images in real-time.
Innovation Solution
The system employs a parallel wavefront matching patch search using hash-based random uniform sampling and acyclic graph computations to determine deterministic matching patch offsets for pixel groups within a target region, enabling consistent and accurate image editing by analyzing pixel groups in parallel and utilizing deterministic search offsets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If randomized searching is used to identify matching patches, then the system can find matching portions of a digital image, but the outputs are unpredictable and non-deterministic
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing hash values of image patches in a database before the actual matching process. This allows the system to deterministically retrieve matching patches without performing randomized searches at runtime, ensuring reproducible results while maintaining ease of operation through efficient database queries.
2Productivity
If the target region is divided into segments for parallel processing, then processing efficiency improves, but accuracy decreases due to inaccurate propagation of matching patch values
Solution Approach 1:
The patent applies segmentation by dividing the target region into multiple segments that can be processed in parallel. Each segment is handled independently with its own matching patch search, allowing efficient utilization of computational resources while maintaining accuracy through independent processing that avoids error propagation between segments.
Solution Approach 2:
The patent introduces an intermediary mechanism in the form of a hierarchical matching approach, where coarse-level matching is performed first to identify candidate regions, followed by fine-level matching within those regions. This intermediary step prevents inaccurate propagation of matching values by establishing proper boundaries and dependencies between different processing levels.
3Manufacturing precision
If the entire target region is analyzed without segmentation, then accuracy improves, but processing time increases exponentially
Solution Approach 1:
The patent applies segmentation to divide the large target region into smaller manageable segments that can be processed independently and in parallel. This maintains the accuracy of analyzing the entire region while reducing processing time by enabling concurrent processing of multiple segments rather than sequential analysis.
Solution Approach 2:
The patent implements partial action by first performing matching patch search on a subset of key segments or representative regions, then using those results to guide or constrain the search in remaining segments. This approach achieves sufficient accuracy for the entire region while significantly reducing total processing time compared to exhaustive analysis.
4Productivity
If parallel processing techniques are used to fill the target region, then processing efficiency improves, but the system cannot produce deterministic results due to random aspects of identifying matching patches
Solution Approach 1:
The patent applies preliminary action by pre-computing hash values and storing them in a database before parallel processing begins. This allows multiple parallel processing threads to deterministically retrieve and compare matching patches based on pre-computed data, ensuring reproducible results while maintaining high processing efficiency through parallel operations.
Solution Approach 2:
The patent uses copying by creating and managing multiple independent copies of the matching patch search process that run in parallel. Each copy operates on its own data structures and produces independent results, allowing deterministic outcomes through controlled replication of the search algorithm across multiple processing units without randomization.
Data Source
AI summary
The present disclosure relates to systems, methods, and non-transitory computer readable media for generating deterministic enhanced digital images based on parallel determinations of pixel group offsets arranged in pixel waves. For example, the disclosed systems can utilize a parallel wave analysis to propagate through pixel groups in a pixel wave of a target region within a digital image to determine matching patch offsets for the pixel groups. The disclosed systems can further utilize the matching patch offsets to generate a deterministic enhanced digital image by filling or replacing pixels of the target region with matching pixels indicated by the matching patch offsets.