enhancement of bone fracture image using filtering techniques
TRANSCRIPT
Enhancement of Bone Fracture Image Using Filtering Techniques Authors: Muhammad Luqman Bin Muhd Zain, Irraivan
Elamvazuthi and Mumtaj Begam Matthew Dunning
Fibula Fracture
Problem & Why It’s Important The issue is that when most ultrasounds are done for
fractures/diseases the picture comes out to be unclear due to speckles in the image.
The importance is that because of these speckles, it does not allow fast interpretations to be made.
Sometimes if fast interpretations cannot be made, if a person has a disease or condition that is life threatening, they cannot get treatment right away.
State of the Art Ultrasounds are non-invasive, portable and do not require
ionizing radiations; however the images are complex. There have been multiple papers produced with different
methods of removing the speckle. Particle Swarm Optimization technique Wavelet Thresholding (Weighted Variance) Novel Bayesian Multiscale Method
A gap based on this paper is that originally, they did not know what type of filtering would be the best result.
Approach to problem The approach to the problem was to use three different
filtering techniques (median, average and Weiner filtering). The original image would go through a contrast
enhancement and then either median filtering, average filtering or a Weiner filter.
The pictures of the images going through each filter were compared for speckle
The data of the image was then shown in a histogram to compare the intensity values, value of pixels
Start
Original Image
Contrast Enhancement
Average FilteringMedian Filtering Weiner Filtering
Result (Output Image)
Histogram
End
Different Filters (Math)
Median Filtering: Used the function medfilt2, to filter the contrasted photo, it replaces each value based upon the neighbors of the value. Great way to remove noise.
Average Filtering: Created a matrix (B) composed of ones, size of image. Then I divided B/size of image. Then I took the contrasted photo and multiplied it by B.
Weiner Filter: restores the image due to a blur, or linear motion.
Contrast: Used function imadjust(picture). Maps grayscale from original image; new matrix
Discussion This method has filled the gap on which method works the
best, it allows the ultrasound to go through a different filter and the results of the histogram show which one does the best job.
Results E) Median reduces speckle noise while keeping edges F) Average reduces speckle and image edges G) Wiener reduces speckle but edges remained intact
Paper discussed that Median and Wiener were close; however Wiener was picked as the best filter because of how its edges remained intact.