|
Regards,
Satips. ![Rose | [Rose]](https://codeproject.freetls.fastly.net/script/Forums/Images/rose.gif)
Don't walk in front of me, I may not follow;
Don't walk behind me, I may not lead;
Walk beside me, and just be my friend. - Albert Camus
|
|
|
|
|
Great
Russell
|
|
|
|
|
I have a set of X-Y-Z coordinate points defining a 3D space (actually contour data). Then supposing I want to find the elevation at point X-Y, would the following method work (or does anyone have any other, better, suggestions):
1) Flatten the XYZ data (i.e., take the X and Y coords)
2) Construct the 2D Delauney triangulation for the data set.
3) Find the triangle in which the point X-Y lies.
4) Use the XYZ coords of the vertices of the triangle to interpolate the elevation at XY.
Of course the hard part is 3, but is the principle sound?
Anyone know of any sample code I can refer to? (bearing in mind I'm not a mathematician or geometrician!)
Cheers
|
|
|
|
|
Years ago when I did some work in this area we had data on a rectangular grid - made things easy
If you google on "gis interpolation" there are a lot of interesting hits,
Peter
"Until the invention of the computer, the machine gun was the device that enabled humans to make the most mistakes in the smallest amount of time."
|
|
|
|
|
I need to retrieve the first four alphanumeric characters from a string, basically,
if I have "DE WINTER", I need to skip the third character and write "DEWI".
Other than going through each substring is there some algorithm for this pattern?
Jon
|
|
|
|
|
It actually depends on the language you're using: string objects usually offer a suitable Replace method.
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
|
|
|
|
|
I'm using JavaScript.
Taking this point of view, the problem is that I need to iterate through each character in the source string and replace non-alphanumeric characters with a blank string ('').
Jon
|
|
|
|
|
This http://www.w3schools.com/jsref/jsref_replace.asp[^]] doesn't help you, does it?
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
|
|
|
|
|
Hiya
i am looking for algorithm for Steganography in .mp3 and .jpg....
I found few links while searching in google but all say something different...So any one here with a good algorithm for this task...
And yeh i want to implement it in C#.Net..Yeh i know there are many articles for stego in this site but they are not for .jpg or .mp3..
thanks
|
|
|
|
|
Search CodeProject for it, you will find several articles.
And yes there are many ways to do it.
|
|
|
|
|
I dont think there is any article for .mp3 and .jpg in code project...Thats what i mentioned in my post...
Yeah there could be many ways, just trying to know which algo is the best way if somebody has already achieved that...
anywayz...
Thanks
|
|
|
|
|
I dont know for mp3, I have seen several for JPEG.
|
|
|
|
|
|
hi !i am writting a project where i have to deal with a lot of matrices and still a biginner in programming.i want to know if there is a library for linear algebra in c and where to get it.thanks!!
|
|
|
|
|
If you're using C++ I'd go for Boost's uBLAS[^] library.
If you are going to use C try the BLAS[^] library.
Steve
|
|
|
|
|
Sure Sir, gonna delete this right away.....
Am a newbie here and wanted to know more on Software Architecture with feedbacks and comments on topic I have raised at my blog pchaitanya.wordpress.com.....
Any feedback would be highly appreciated....
Thanks,
Chaitanya
Chaitanya
|
|
|
|
|
If the Lord God Almighty had consulted me before embarking upon the Creation, I would have recommended something simpler.
-- Alfonso the Wise, 13th Century King of Castile.
|
|
|
|
|
For a 2-dimensional plane, I've implemented a function for determining whether a circle intersects a ray. Suppose there is one ray and many, many circles. How would I reduce the number of tests by selecting the circles that are certain or likely to be hit?
ROFLOLMFAO
|
|
|
|
|
Hi,
lots can be done depending on circumstances.
SCHEME 1
Assuming you are only interested in a finite and rectangular part of the 2D plane,
you could divide the area in a number of smaller rectangles. Say 16*16 rectangles.
Now a ray would intersect only some of these rectangles, so we are only
interested in those circles that partially cover those rectangles.
Assuming your circles are static, that can be precalculated and stored: for
each record, you can hold a list of crossing circles.
So now it is a matter of finding the rectangles crossed, then scanning the circles
listed for these rectangles.
WARNING: things must be organized in such a way that the same circle is not
tested again for different rectangles; a boolean flag might help.
SCHEME 2
Assuming (most of) the circles are fairly small with respect to the entire area,
if the ray does not cross the entire area, but is restricted to some X-range
(for a steep ray) or some Y-range, one could easily eliminate all circles that
lie fully outside said X or Y-range.
Hope this helps.
|
|
|
|
|
After some more thought about the problem, I realized that while we can't elimenate unnecessary tests altogether, we can cut out a good portion of the tests in some circumstances. This might involve maintaining certain properties before the actual hit testing occurs (like how order is maintained for a balanced search tree before a search is performed).
I will outline my approach below (and you can judge my methods):
- A rectangular plane (named: MAP) is created which everything else will be relative to.
- MAP is subdivided into regions. These regions are squares. They contain references to circles. The length and width of the region are set such that it is not smaller than the largest diameter of a circle. These regions are addressed by indices. The indices are directly related to the region's offset from the origin on MAP.
- A circle (named: CHAR1) is created on MAP with a radius no greater than half a region's length. (This is important later on when we try to determine what regions get tested.)
- CHAR1 is added to the region that it is in. (The circle's boundaries may cross into one or more other regions.) Its region will now have a reference to it.
- CHAR1 moves and it crosses its region's boundaries. It has crossed into another region, so the region's references to CHAR1 are deleted, and the current region CHAR1 resides in creates a reference to CHAR1. (Regions are synchronized with the circle's location.)
- Another circle (named: CHAR2) is created on MAP. It fires a projectile that follows a line (named: PATH).
- Now here is the tricky part (I don't even know if it works.):
- With PATH defined by y = mx + b, create 2 more lines that are parrallel to PATH and are separated by an a perpendicular distance that is equal to the greatest possible radius of a circle (a limit which is artificially imposed). The lines are created on with sides of PATH. The three lines intersect all the regions that should be tested. (Hit-testing with squares should be much more efficient than hit testing with circles.)
- So now that we have 3 lines, we just plugin values for X in increments of the width of the region starting from the position of CHAR2. We get coordinates that hit every region that needs to be tested. Suppose the indices of the regions use the format [x, y] in increments of 1, and region's corner defined by it's coordinates multiplied by the region length/width, then we could do an integer division on the coordinates given by plugging in X values into the line equations and get the regions that should be tested.
ROFLOLMFAO
|
|
|
|
|
Hi,
this sounds reasonable, provided:
1) there are many CHAR1 (otherwise it all does not matter)
2) the greatest possible radius of a circle is small with respect to MAP size
I am not sure about your integer division stuff; the line y = mx + b seems to
imply floating-point: m and b wont be integers.
BTW: the general equation for a line is a*x + b*y + c = 0 which is symmetric in
x and y, resulting in no trouble when the line is vertical (if you dont like having
3 coefficients, you can divide them all by c provided c is non-zero).
Further thought: there is a useful method Region.IsVisible()
you might apply it to a region that is a (non-orthogonal) rectangle starting at
the shooter, pointing in the direction he is shooting, and having a width of
twice "the greatest possible radius of a circle" and a length sufficient to
cross the border of MAP.
Using it you would mean you do not need your squares after all, simply test each
circle's center against it (very simple, probably fast too).
As a side effect you could shoot circles that are off-screen too (by making the
rectangle length quite large) !
|
|
|
|
|
Thanks for your responses. I'll try to work with different approaches to fidn the best one. (It's a Windows Presentation Foundation application by the way. It's also network-oriented. The game engine runs on the server.)
As for the integer division stuff, I'll probably cast the Double to an Int64, strip the fractional part, or round down after division — whichever is more efficient.
ROFLOLMFAO
|
|
|
|
|
Just to ensure that you know that testing for intersection with circles is reasonably inexpensive:
if floating point, normalise equation of line
ax + by + c = 0
so that a^2+b^2=1
then the test for intersection with circle centre (x0,y0) radius r0 is
abs(a*x0 + b*y0 + c) <= r0
If integers, then there are several ways, but
(a*x0 + b*y0 +c)^2 < (a^2+b^2)*r0^2
is reasonably cheap (4 multiplications and 2 additions) assuming that r0^2 and (a^2+b^2) is precomputed. You may have to be careful about overflows.
With modern GFLOP processors, you can test a lot of circles in real time. It all depends on how many circles you have and how they are organised.
Peter
"Until the invention of the computer, the machine gun was the device that enabled humans to make the most mistakes in the smallest amount of time."
|
|
|
|
|
Hi,
this is great, and mathematically correct.
When all coordinates use no more than 14 bits (that is 16000 pixels), then everything
can be calculated in 32-bit signed integer arithmetic without risking overflows.
Seems much better than the sorted-squares stuff...
|
|
|
|
|
Hmm… I'd have to agree that dividing MAP into regions would be horribly complex (I haven't even implemented the code yet), but then again, I'm restricted to floating-point math, so I can't take the integer math approach. I'm also quite skeptical about testing every circle. Maybe I should make it more scalable, and allow automatic selection of algorithms based on certain conditions.
In the end, we converge towards an optimal solution, but we never reach it.
ROFLOLMFAO
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
