Presentation is loading. Please wait.

Presentation is loading. Please wait.

Levi Smith REU Week 1.

Similar presentations


Presentation on theme: "Levi Smith REU Week 1."— Presentation transcript:

1 Levi Smith REU Week 1

2 Challenges Being almost completely new to computer vision, these are a lot of new concepts to take in. Getting used to programming in Matlab Working out the finer details of the algorithms involved when it comes to actually implementing the techniques.

3 Edge detection Roberts Edge Detector
Implemented the Roberts Edge detector on this image. Computed the horizontal edges in all three individual color channels. Combined them into one single image to display differences and similarities between the edges in different colors. Edge detection

4 Edge Detection Sobel Edge Detector
Implemented the Sobel Edge detector on this image. Once again, computed the horizontal edges in all three individual color channels. Also combined them into one single image to display differences and similarities between the edges in different colors. Edge Detection

5 Implemented a function that computes the Laplacian of Gaussian of an image.
This function takes advantage of the property of separability of Laplacian of Gaussian by computing it with two one- dimensional filters rather than one two-dimensional filter. Laplacian of Gaussian

6 Computed the Canny Edge detection for the following image.
Computed the magnitude and direction of the gradient of the image. However, was not able to get working non-maximal suppression or the hysteresis threshold in my implementation. Canny Edge Detector

7 Harris Corner Detector
Computed points of interest in this image using the Harris Corner Detector. Again, had difficulty getting a working implementation, but was able to achieve results that highlight key points of interest. I then added a threshold to the computed corner points to filter out unwanted edges and flat regions. Harris Corner Detector

8 Dense Sift My implementation of the dense SIFT had good results when using the same image rotated by an angle of 90 degrees.

9 Dense Sift On the other hand. It did not have good results when the image was rotated by an angle not an increment of 90 degrees.


Download ppt "Levi Smith REU Week 1."

Similar presentations


Ads by Google