Blur Detection in Digital Image Scanning via Patch-Based Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-throughput scanning of histological slides faces bottlenecks in quality control due to manual detection of out-of-focus regions and defects, which is time-consuming and inefficient.
Innovation Solution
A system utilizing processors to generate patches from digital images, calculate sharpness metrics, determine blur scores, and create blur maps using algorithms like random forest regression or residual neural networks to automatically identify and flag blurry areas, thereby streamlining quality control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual screening is used to detect out of focus regions and defects, then measurement precision can be maintained, but productivity is reduced due to time-consuming manual detection
Solution Approach 1:
The patent replaces manual mechanical screening with an automated computational system that uses image processing algorithms to detect blur and defects. The system calculates sharpness metrics (gradient-based, frequency-domain, wavelet transforms) and classifies patches to automatically identify out-of-focus regions, eliminating the need for manual visual inspection while maintaining detection accuracy.
Solution Approach 2:
The patent creates a digital copy of the manual inspection process by training a classification system on labeled data. The system learns from annotated examples of blurry and sharp regions, then applies this knowledge to automatically screen new images. This copying of the expert inspector's judgment enables automated high-throughput screening that maintains the precision of manual detection.
2Productivity
If automated blur detection is implemented, then productivity is improved, but device complexity increases due to multiple processing stages
Solution Approach 1:
The patent divides the image into multiple patches and applies different sharpness metrics to each patch independently. This segmentation allows the system to process large images efficiently by treating them as collections of smaller units, each evaluated by appropriate metrics (gradient-based, frequency-domain, wavelet). The modular patch-based approach simplifies the overall complexity by breaking down the complex task of full-image blur detection into manageable sub-tasks.
Solution Approach 2:
The patent employs multiple sharpness metrics that operate in different parameter spaces (spatial domain, frequency domain, wavelet domain). By transforming the image data into different representations and applying specialized metrics to each, the system achieves robust blur detection without requiring a single overly complex algorithm. Each metric operates with relatively simple computations, and their combination provides comprehensive coverage.
3Measurement precision
If multiple sharpness metrics are calculated for each patch, then measurement precision is improved, but use of energy increases due to computational requirements
Solution Approach 1:
The patent calculates multiple sharpness metrics for each patch, which can be considered excessive action. By computing gradient-based metrics, frequency-domain metrics, wavelet metrics, and other sharpness measures simultaneously, the system over-samples the feature space to ensure high detection accuracy. This partial application of multiple metrics to each patch (rather than applying all possible metrics to the entire image) balances computational energy consumption with the need for precise blur detection.
Data Source
AI summary
The present disclosure discusses systems and methods to detect blur in digital images. The solution can be incorporated into the quality control systems of pathology and other slide scanners or can be a stand-alone solution. The solution can identify scanned images that include blur and cause the scanner to automatically rescan the blurry image. The solution can also identify regions of the scanned image that include blur. The solution can generate blur maps for each of the scanned images that identify regions of the scanned image that include blur.