Computer Vision - Assignment 4

Gabor Filters

Visualize texture image after filtering using multiple Gabor Filters as well as the Kernel of filter itself.

Discuss the parameters chosen to create Gabor filter and compare with the resulting image after filtering.

Choose an image from the following texture dataset: link

functions: getGaborKernel

Hu Moments - Shape features

Create your own image that will show 3 different hand gestures (for example from a video)

Binarize the image to create a mask of the hand gesture
- Compute and visualize contours of your gesture
Compute HuMoments from the created mask image / contours for each gesture
- Show computed values for each gesture

Create a simple program using your camera that will check the current hand gesture on camera and compare it with the trained image / HuMoments features you learned earlier to output the resulting gesture on the screen. Experiment with different distance metrics. Implement a simple threshold value based on the distance metric in order to filter out unknown or new gestures.

Object Description and Detection

Use your knowledge of local descriptors to detect and localize image patch within the larger image from following data - link
- Experiment with feature detectors: SIFT, FAST, Harris
- Experiment with feature descriptors: SIFT, SURF, ORB
- Compute homography matrix and localize your image patch (RANSAC)
  - Find the best combination of detection & description for your task

Report abuse