Blood Vessel Image Sharpening via Gabor Filtering and Edge Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for enhancing blood vessel image quality, such as infrared imaging, face challenges in accurately distinguishing between blood vessel and non-blood vessel domains due to rotation, movement, and distortion, leading to inaccurate separation and prolonged processing times.
Innovation Solution
A method involving Gabor filtering to sharpen blood vessel images based on various directions and thicknesses, followed by edge detection using first-order or second-order differentiation, and neural network-based fusion to improve image clarity through score level fusion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If infrared imaging method is used to obtain blood vessel images, then the method is less objectionable, inexpensive and has no side effects, but the blood vessel and non-blood vessel domains cannot be clearly distinguished
Solution Approach 1:
The patent combines multiple image processing techniques including Gabor filtering, edge detection, and neural network-based fusion to process infrared blood vessel images. By merging these different processing stages, the system enhances the clarity of blood vessel domains while maintaining the advantages of infrared imaging (low cost, no side effects), thus resolving the contradiction between ease of manufacture and measurement precision.
Solution Approach 2:
The patent applies Gabor filtering with multiple directions and scales to enhance blood vessel images. By changing the parameters of the filter (directions, scales, and thresholds), the system optimizes the enhancement of blood vessel signals while suppressing background noise, thereby improving image quality without changing the infrared imaging method itself.
2Reliability
If matched filter, Wiener filter and average filter are used to remove noise, then noise is reduced, but the processed image becomes out of focus and blood vessel separation becomes inaccurate
Solution Approach 1:
Instead of applying traditional noise reduction filters that blur the image, the patent extracts edges from the noisy infrared image using edge detection algorithms. By taking out only the essential edge information and using neural networks to fuse this extracted information with the original image, the system reduces noise while maintaining image focus and separation accuracy.
3Measurement precision
If two high dynamic range images at different exposure times are used for sharpening, then image sharpening is achieved, but rotation and movement between images cause matching difficulties and prolonged processing time
Solution Approach 1:
The patent performs edge detection and neural network-based fusion on a single infrared blood vessel image without requiring multiple images. By taking preliminary actions (edge extraction, filtering, and fusion) on one image, the system achieves sharpening效果 without the time-consuming process of matching and aligning multiple images, thus resolving the contradiction between sharpening quality and processing time.
Data Source
AI summary
A method of processing a blood vessel image is provided. The method includes (a) sharpening an original blood vessel image using a Gabor filter in consideration of various directions and thicknesses of blood vessels included in the blood vessel and (b) detecting edges according to a change in brightness in a blood vessel domain and a non-blood vessel domain of the original blood vessel image and the blood vessel image on which the Gabor filtering step is completed, using an edge extraction method based on a first-order differentiation or second-order differentiation.