Matching method for laser radar range image with moment invariants

In order to study the laser radar range image matching with scale and rotation, a set of matching methods was introduced. It included adjustment for images proportion, extraction of the Krawtchouk moment invariants, formation of the feature space and image searching and matching. With the simulated laser radar range image for experimental data, by adjusting the proportionately of the reference image and real-time image and extracting their Krawtchouk moment invariants to format feature vection, the simulated real-time laser radar range image was successfully matched with the simulated reference laser radar range image. Experimental results show that Krawtchouk moment invariants are suitable for digital laser radar range image matching. This approach is easy and convenient. Furthermore, a desirable matching result can be obtained with a certain proportion of scaling and rotation at any angle and translation at any degree distortion of the image. This algorithm has strong ability to overcome Gaussian noise with high matching and location accuracy.