Thread Tools |
This thread is privately moderated by Jack Crossfire, who may elect to delete unwanted replies. |
|
|
|
Discussion
6.5 mile RC car drive
A test of SURF image matching with different focal lengths & lighting was a complete failure.
So did 4 x 8 minute miles on the same section of track, gathering 2 runs in both directions, in equivalent lighting & focal length, high quality JPEG. Wanted to do more repeats, but sunburning became an issue for the human. The old raspberry pi was still putting out 10fps. Gave up trying to configure it over wifi. The USB port has enough power for a webcam & wifi. The picture quality was still awful. Stepping up the saturation revealed it was probably using an analog NTSC signal in 1 stage. The proprietary Logitec chip still takes in a digital signal. The method for comparing features was a very old source code example called "naive nearest neighbor". It too was pretty bad. Source code examples of using OpenCV for image matching are few & far between & they all interface the library at different points. The image matching examples seem to base the matched image on the number of matching features & only vary the algorithm used to match the features. In the path database, most sections of the database had the same number of matching features for all positions. The theory was always that there would be matches between the current image & large swaths of the database. It would need another step. The image with the features closest in scale & relative positions to the current image would be the current position in the database. Setting the frames in motion revealed most of the matched features were indeed hugging a set of trees which was indeed changing size. There was a lot of noise which was to be expected. Unfortunately, all the trees are dying off & being chopped down due to lack of water. |
|
|
|
|