CT Fiber Image Binarization via HSV Tangent Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current binarization methods for CT sectional images of fiber packages struggle to effectively separate the optical fiber core from a complex background with similar gray scales and are prone to artifacts, leading to suboptimal identification and reconstruction simulations.
Innovation Solution
A binarization method involving HSV image model conversion, nonlinear brightness adjustment using a composite tangent function, morphological structural element deletion, gray value interval expansion, and local adaptive threshold binarization to enhance image contrast and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If global threshold binarization is used, then processing speed is fast and algorithm is simple, but it cannot handle complex situations like uneven illumination and artifacts
Solution Approach 1:
The patent divides the image processing into multiple stages: first applying global thresholding for initial binarization, then identifying artifact regions, and finally applying local thresholding only to artifact regions. This segmentation approach maintains the speed advantage of global thresholding while improving accuracy in problematic areas through targeted local processing.
Solution Approach 2:
The patent applies different binarization strategies to different regions of the image. Global thresholding is applied to normal regions, while local adaptive thresholding is applied specifically to artifact regions identified through image analysis. This local quality approach ensures optimal processing for each region's characteristics.
2Reliability
If local threshold binarization is used, then it can adapt to complex situations like uneven illumination, but it causes small background pieces to be binarized into foreground, resulting in more noise
Solution Approach 1:
The patent extracts and isolates artifact regions from the rest of the image through specific image processing steps. By identifying and separating artifact regions, the method applies local thresholding only where needed, preventing the noise generation problem that occurs when local thresholding is applied globally to entire images.
Solution Approach 2:
Instead of applying local thresholding to the entire image, the patent applies it partially only to identified artifact regions. This partial action approach maintains the benefits of local adaptive thresholding in problematic areas while avoiding the noise generation that would occur if local thresholding were applied excessively to the whole image.
3Device complexity
If existing binarization methods are used on CT images with artifacts, then processing is simple, but the target object cannot be effectively separated from background with similar gray scales
Solution Approach 1:
The patent performs preliminary analysis to identify artifact regions before applying the final binarization strategy. By detecting artifact regions in advance and preparing appropriate processing methods for these regions, the system improves target object separation accuracy without significantly increasing overall processing complexity.
Solution Approach 2:
The patent introduces an intermediary step of artifact region detection and classification between the initial image input and final binarization. This intermediary process identifies problematic regions and directs appropriate processing methods, enabling accurate target object separation while maintaining reasonable processing complexity.
Data Source
AI summary
Provided is a binarization method for CT sectional image of fiber package containing artifacts, which includes: performing brightness adjustment on the source image obtained after converting of the HSV image model by using a composite tangent function; creating a planar morphological structural element having a morphology similar to that of a target object to obtain a background image without the target object; obtaining a second intermediate image by a subtraction operation of the first intermediate image and the background image; improving an image contrast of the second intermediate image again to obtain a third intermediate image; and binarizing the third intermediate image by using a local adaptive threshold binarization algorithm and removing a background noise to obtain a final binarized image. The binarization method can improve the uneven brightness of the image under complex illumination, alleviate the artifacts, and strip similar objects from the background with similar gray scales.