All entries for Monday 13 November 2006

Delegation of Work

Follow-up to C++ code so far from DCS Robot Vision System

OK guys, after a lot of deliberating, i think that the best way for us to split up the work is for each of us to convert one of the following classes into c++:

ImageProcessor – Me
Histogram – Hoi (so you can also add a histogram for the contourprocessor)
ColourMask
ContourProcessor
EdgeDetector

Remaining people if you could comment on this entry with which you will be doing, then we wont end up doing the same thing twice!

In regards to the formatting of image throughout processing, it will be easiest to stick with CImg, and we still need to decide how we will get around the RGB colour issue with the CImg library, as I can’t seem to get it to produce the RGB values of each pixel, and have so far been doing it based on percentages of colour in each pixel.
Let us all know if one of you finds a way around this. Otherwise for all methods assume the same types for parameters and returns, and as i said use CImg rather than NCCTable.

Any questions just post them up!!!

T

: 13 Nov 2006 16:43 |  Comments (6) |  Close comments |  Report a problem

Question about code

Guys,
Which parts of the java do we NOT need to convert to c++?

Is it just the FrameAccess and JpegImagesToMovie classes?

This would leave:
ImageProcessor
ColourMask
ContourProcessor
EdgeDetector
Histogram

which definitely need converting, and depending on how we are going to store the image and colour values while we are processing, we may need equivalent classes to:
NCCTable
Pixelizer

Just let me know your thoughts….

Cheers,
T

: 13 Nov 2006 12:45 |  Comments (1) |  Close comments |  Report a problem

C++ code so far

Here is the C++ code so far.

Code

The part which I have done is in normalize.cpp, where I have basically added to Ants code.

I will sort out what needs to be done before lunch today, and post on here or email you guys with the equivalent java method, and spec for the c++ method which needs to be created.

Hopefully we can get all this done by Wednesday, then we can start testing it on the Robot by the end of the week so we can have some preparation time for the presentation over the weekend.

: 13 Nov 2006 10:20 |  Comments (0) |  Close comments |  Report a problem

Code

Follow-up to Can we stop developing in Java? from DCS Robot Vision System

Absolutely, the reason why I continued using Java is because I had no base to use to continue the work in c++ . Spending a day on it in dcs would be a very good idea, plus it’ll help speed up us getting to know C++ and CImg better.

We can discuess how we’re going to sort out the work load tomorrow. I’ll stop doing any more updates to the Java code now, but before I do I just want to post the newest version. It implements the contour recognition stuff, and joins the JpegImagesToMovie class with the main program.

Remember that we don’t need to implement a C++ version of the FrameAccess or JpegImagesToMovie files, so don’t stress about those. We can use the java ones to produce a video/image demonstration in the presentation.

Code here

: 13 Nov 2006 01:09 |  Comments (0) |  Close comments |  Report a problem