Feature detection and matching example

This is an example to show how feature point detection can be used to find a registered planar object from video images.

Registration step:

Detection step:

The number of matching is not enough in the above example image, but it may help you find how it works.

Prerequisite

Python 3.5 and OpenCV 3.

1. Download and install Anaconda3-4.2 (64bit version, python 3.5) from https://repo.continuum.io/archive/. (Note: Not the latest version)

2. Open Anaconda Prompt as administrator.

3. Install OpenCV.

   > conda install -c menpo opencv3 #Should be "menpo" instead of "mempo"
   

How to use

1. Run the example code.

   > python object_detection.py
   

2. Press 'r' to register a query image with a target (textured planar) object. Then object detection will start.

3. Press 'q' to quit the detection.

Tips: Use high-contrast (e.g., black and white) textured planar object.

Bug

Error check is not enough and the program may be terminated at for example perspectiveTransform in some cases.