Image Processing and New OEM GPS Tutorial

I just published a new GPS tutorial over at Instructables. I wanted to demonstrate how easy it is to hook up an OEM GPS unit instead of having to fool with a commercial model. I also entered the tutorial into the Instructables contest for RoboGames. If you like or find the tutorial helpful, I would appreciate it if you would vote for me! (The second image on this post is of the setup in the corner of my parents basement I used to make the video.)

I have spent my time since the last post continuing to elaborate on the flash security system. It will be deployed for real world use this week. I have also been working on neural networks in c++ and using qt in order to load images. I have also spent the last few days working on doing image capture and compression inside of flash, then sending that to the server via RTMP and AMF3 only. It looks like before AMF3 and flex the only way to send images was by using an additional scripting language in combination with flash. There are a ton of tutorials about how to do it this way, or use ffmpeg on the server side to capture images. However, I think the way I am doing it should prove to be a little more optimal as it just uses Red5(Java) on the server and Flash on the client. Nothing else is needed.

The method I am using takes a bitmapimage from the video stream on the flash client, encodes it to a png or jpeg, then makes that a bytearray, then sends that to the server using a netconnection.call. On the server I throw that data into a byte array, then use java ImageIO in order to re-construct the bytearray and save the stream capture to an image on the server. I am pretty sure you could do this process in reverse in order to provide stream captures of a video to the client prior to them streaming it. If I have time I may give this a try and post the results to the Red5 mailing list.

Lastly, I wanted to quickly put a few links to two papers that have helped me understand the current state of vision processing much better. The first is a thesis on the Phission Vision Processing system called "Phission: A Concurrent Vision Processing System Software Development Kit for Mobile Robots" by Philip D.S. Thoren. The second is "Vision for Mobile Robot Navigation: A Survey" by Guilherme N. DeSouza and Avinash C. Kak. I am unable to find a suitable link for the second at the moment, but I know it's available. I won't go in to it here, but lets just say I strongly oppose the hoarding of journals by places like IEEE and ACM. I have always been a huge fan of the DOAJ.