Mars Project

A computer vision project to extract features from a single dense image of Mars from the Mars Reconnaissance Orbiter.

Project Plan

1. Examine image for extractable features
   a. Crater count [start here]
   i. SkImage example
   b. Crater diameters (requires extra data)
   c. Ejecta diameters
   d. Ridge counts
   e. Compare smooth to cratered surface statistics
   f. Boulder counts

2. Determine package and pre-trained model to use a. Python
   i. OpenCV
   ii. Scikit-Image
   a. https://scikit-image.org/docs/stable/user_guide/
   iii. PIL (Python Imaging Library)
   iv. Tensorflow
   v. Keras
   b. R
   i. Rvision

3. Determine computer vision model to use
   a. Gabor filter banks for texture classification
   b. Local Binary Pattern for texture classification
   c. Multi-Block Local Binary Pattern for texture classification
   d. Morphological Filtering
   e. Colocalization metrics
   f. Registration using optical flow
   g. Removing small objects in grayscale images with a top hat filter
   h. Using window functions with images

Files

1. ESP_072719_1970_RED.browse.jpg is the low-resolution dataset preview image.
2. ESP_072719_1970_RED.LBL has the detailed information about the image in the LBL format.
3. ESP_072719_1970_RED.XML has the detailed information about the image in the XML format.

Data Source

1. Mars Reconnaissance Orbiter ESP_072719_1970_RED.JP2

Other Notes

Consider combining algorithms. E.g. hole detection with blob detection