Edge Sharpness Estimation via Grayscale Slope Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating sharpness in images, particularly for document pages, are inadequate as they rely on measures developed for natural scenes and require a reference image or are coarse, failing to accurately assess blurriness in real-time on mobile devices.
Innovation Solution
The proposed solution involves a method that captures images and generates smoothed images in both x and y directions, identifying windows around edge pixels to estimate edge sharpness by determining differences in grayscale values, using the maxDoM and ΔDoM models to quantify edge width and slope changes, which are then combined to assess overall image sharpness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If related art no-reference methods are applied to all pixels using statistical characteristics, then sharpness estimation can be performed without a reference image, but the measurement precision is insufficient for document images with high contrast transitions
Solution Approach 1:
The patent segments the image processing by identifying edge pixels first, then applying different processing to edge regions versus non-edge regions. Edge pixels are detected using gradient magnitude thresholds, and a window around each edge pixel is analyzed separately. This segmentation allows precise measurement of edge sharpness while avoiding noise from non-edge areas, directly resolving the contradiction between no-reference capability and measurement precision.
Solution Approach 2:
The patent applies local quality analysis by computing sharpness metrics specifically at edge locations rather than uniformly across all pixels. A window of pixels surrounding each edge pixel is analyzed to compute local gradient statistics. This localized approach provides high precision for document images where sharpness is concentrated at text edges, while maintaining no-reference operation by using only local pixel values.
2Ease of manufacture
If edge width is measured by counting pixels with increasing/decreasing grayscale values, then a simple no-reference method can be implemented, but the measurement precision is coarse and quantized to pixel widths
Solution Approach 1:
The patent changes the measurement parameter from discrete pixel counting to continuous gradient magnitude analysis. Instead of counting pixels with increasing/decreasing grayscale values, the method computes gradient magnitudes and uses statistical properties (mean, standard deviation) of these gradient values within a window. This parameter change maintains computational simplicity while providing sub-pixel precision for edge width measurement, resolving the contradiction between ease of implementation and measurement precision.
3Measurement precision
If related art methods require a perfect reference image for comparison, then measurement precision can be high for evaluating degradation, but the device complexity and ease of operation are reduced due to the need for reference image acquisition and alignment
Solution Approach 1:
The patent implements self-service by enabling the sharpness estimation system to evaluate image quality using only the image itself, without requiring an external reference image. The method computes gradient statistics and edge sharpness metrics directly from the input image's pixel values and their local variations. This self-contained approach maintains high measurement precision for document images while eliminating the complexity of reference image acquisition, storage, and alignment systems.
Data Source
AI summary
Systems and methods described herein are directed to estimating the sharpness of images or photos such as document pages or scenes. The systems and methods are based on calculating a difference in grayscale values between pixels at an edge of an image. The differences may be accomplished by taking a slope of grayscale values between two pixels at an edge window, and estimating the edge sharpness based on the slope, multiple slopes, or differences in the slopes.