Foreground-Background Segmentation via Background Surface Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Camera-captured document images often suffer from uneven illumination and blurriness, making it difficult to segment foreground and background effectively, which hampers optical character recognition and document digitization.
Innovation Solution
A system and method that applies a watershed transform to the noise-thresholded color gradient magnitude, estimates the background surface, normalizes the image, and uses Euclidian color distance to separate foreground and background regions, reducing over-segmentation and improving image clarity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If global thresholding is used for segmentation, then processing speed is fast, but segmentation accuracy deteriorates under non-uniform lighting
Solution Approach 1:
The patent applies adaptive thresholding that computes different threshold values for different regions of the image based on local statistics. Each pixel's threshold is determined by its neighborhood characteristics, allowing the segmentation to adapt to local lighting variations while maintaining overall processing efficiency.
Solution Approach 2:
The patent dynamically adjusts threshold parameters based on local image statistics such as mean and standard deviation of pixel intensities in neighboring regions. This parameter adaptation allows the segmentation algorithm to respond to varying lighting conditions without requiring manual intervention.
2Measurement precision
If adaptive thresholding is used for segmentation, then segmentation accuracy improves under non-uniform lighting, but device complexity increases
Solution Approach 1:
The patent divides the image into local neighborhoods or blocks and computes threshold values independently for each region. This segmentation approach allows complex adaptive processing to be applied locally rather than globally, reducing computational overhead while maintaining accuracy.
Solution Approach 2:
The patent applies adaptive thresholding selectively to regions where it is most needed (areas with significant lighting variations) rather than uniformly across the entire image. This partial application reduces computational complexity while maintaining segmentation accuracy in critical regions.
3Ease of operation
If camera is used for document capture, then portability and ease of operation improve, but image quality deteriorates due to movement and uneven lighting
Solution Approach 1:
The patent extracts and removes the effects of uneven lighting and camera movement artifacts from the captured image through background modeling and foreground extraction techniques. By separating the document content from the degraded background, the method recovers high-quality text and image data.
Solution Approach 2:
The patent introduces intermediate processing steps including background estimation, gradient computation, and morphological operations that act as mediators between the raw degraded image and the final segmented result. These intermediate transformations progressively improve image quality while preserving the benefits of portable camera capture.
Data Source
AI summary
A system and method for segmenting foreground and background regions on a digitized image uses a computer, having a processor and system memory, to segment the image into initial regions and identify background regions from the initial regions. A complete background surface is estimated of the image, and pixels of the image are rectified with the estimated background surface to normalize the image. Normalized pixels are compared with a threshold color to determine a final segmentation of background regions.