Flood Fill and Floor Finder algorithm?
Trung from Vietnam  [4 posts]
9 years
 Based on obstacle avoidance tutorial, I'm trying to simulate two modules Floodfill and Floor Finder in Matlab, but I can't find the algorithm to implement them.Does anyone know about these algorithms?Please help me.Thanks.
Anonymous 9 years
 These algorithm exist in RR see http://www.roborealm.com/help/Flood_Fill.php and http://www.roborealm.com/help/Floor_Finder.php. You will need to utilize the modules within RR and then send results to Matlab fpr futher processing. Download an example dll for Matlab and RR at http://www.roborealm.com/downloads/API.zip
Trung from Vietnam  [4 posts] 9 years
 The problem is that I can't get the idea from the instruction of those 2 modules since they looks like a short introduction than showing an algorithm.I really need more detail explanation or a clearer algorithm to follow.I intend to use Matlab independently to process the image.Could anyone help me please?Thanks
Anonymous from United Kingdom  [99 posts] 9 years
 First you should take a look at this for a general overview on different techniques for image segmentation. http://en.wikipedia.org/wiki/Segmentation_(image_processing) I am not sure exactly how STeven implemented the floor finding algorithm, but I would guess it is a variant on some sort of Region Growing Method.  Basically you take a set of seeds (The sampled region in this cased)  You compare the intensity of neighboring pixels to the the seeds.  If they match within the given threshold then they are still part of the floor, if not, then a new region would be created,  since the floor finder module is only interested in creating one region, it can dump all regions that are not considered attached to the initial seeds. Hope that makes some amount of sense.  The original technique was described by haralick and shapiro.  They put out a book pretty recently on Computer and Robotic Vision (1&2)  Any text on computer vision should go through their algorithm (or maybe I gave you enough to get started?)  If you want more, you might look at: http://www.cs.cmu.edu/~cil/v-source.html STeven will come along in a week or so and sort this all out.
Anonymous from United Kingdom  [99 posts] 9 years
 I just read a nice paper on this tonight.   DeSouza and Kak "Vision for Mobile Robot Navigation. A Survey" http://www.societyofrobots.com/robottheory/Survey_of_vision-based_robot_control.pdf
Anonymous 9 years
 FloodFill is just your generic color segmentation algorithm. You should be able to find the actual code in one of the open source systems. See http://www.roborealm.com/links/vision_software.php for a list. Basically it just groups like colors (RGB values) with each other given a certain threshold. The flood finder is similar. Use the sample pixels in front of the robot and extrapolate the same colors out to the rest of the image. There have been several papers that use/introduce this technique so it is quite well know. Check researchindex.org or google for those papers. They are not hidden or propriety secrets but well known techniques in the vision robotics community. STeven.
Trung from Vietnam  [1 posts] 9 years
 Hi,Steven What is the name of the technique which refer to the floor finder module? Trung.
Anonymous 9 years
 Search for "histogram segmentation". STeven.

 This forum thread has been closed due to inactivity (more than 4 months) or number of replies (more than 50 messages). Please start a New Post and enter a new forum thread with the appropriate title.