RoboSub Club of the Palouse

Washington State University & University of Idaho

Computer Vision Matching Process

The image matching process

The image matching process

The submarine is required to recognize many obstacles throughout the competition. As a result the sub must have image recognition and matching capabilities. Our sub has two cameras, one facing forward and one facing down. The images recorded from each camera continuously go through the process displayed in the diagram of recording and image then analysing it.

Our analysing process starts by looking for the known colors of obstacles in the competition. When they are found it records those sections of the image and analyses them for the known shapes of the obstacles of that color. If a match is found then the name of the obstacle, it’s location, size and orientation are sent to the artificial intelligence program to decide what needs to be done. The process is then repeated.

Your email address will not be published. Required fields are marked *

*