I am not sure if it is proper to ask for help on algorithm here, but could anyone give me some guide, or just tell me where I could find such a kind of guide? Thanks a lot!
The problem is like this: given a fixed number of circles, I need an algorithm to find an optimal set of positions and radius of these circles to cover a given shape, so the error area (the parts of the circles outside the given shape + the parts of the shape not covered by these circles) is minimal? Circles could overlap.
This is not a trivial problem and there is certainly no simple analytic solution. For example: even the simplest version - one circle and one simple connected area isn't necessarily easy to solve depending on the shape of the area. There will also typically be numerous false minimums.
I would suggest that simulated annealing would be a suitable technique to find a good (if not the optimal) solution. Effectively, with n circles you are exploring a wildly varying function of 3n variables (x, y, and r for each circle) and simulated annealing is a fairly efficient way of exploring such an environment.
Related
I have a few points randomly distributed over a 2D-map. I also have a finite number of circles that I want to place so they cover as many of the points as possible, kind of like a turret-game AI that places turrets in a base to protect valuable buildings. Is there any good way to do this?
What you are describing sounds like a form of the maximum coverage problem. One simple way to solve this problem is applying the greedy algorithm.
This means you start by drawing the first circle such that it covers the biggest possible region. Then you draw the second circle such that it covers the biggest possible area and so on.
BACKGROUND
So I'm creating a program that recognizes chess moves. So far, I have implemented a fair number of algorithms to come up with the best results possible. What I've found so far is that the combination of undistorting an image (using undistort ), then applying a histogram equalization algorithm, and finally the goodFeaturesToTrack algorithm (I've found this to be better than the harris corner detection) yields pretty decent results. The goal here is to have every corner of every square accounted for with a point. That way, when I apply canny edge detection, I can process individual squares.
EXAMPLE
WHAT I'VE CONSIDERED
http://www.nandanbanerjee.com/index.php?option=com_content&view=article&id=71:buttercup-chess-robot&catid=78&Itemid=470
To summarize the link above, the idea is to find the upper-leftmost, upper-rightmost, lower-leftmost, and lower-rightmost points and divide the distance between them by eight. From there you would come up with probable points and compare them to the points that are actually on the board. If one of the points doesn't match, simply replace the point.
I've also considered some sort of mode, like finding the distance between neighboring points and storing them in a list. Then I would perform a mode operation to figure out the most probable distance and use that to draw points.
QUESTION
As you can see, the points are fairly accurate over most of the squares (though there are random points that do not do what I want). My question is what do you think the best way to find all corners on the chessboard (I'm open to all ideas) and could you give me a somewhat detailed description (just enough to steer me in the right direction or more if you choose :)? Also, (and this is a secondary question) do you have any recommendations on how to proceed in order to best recognize a move? I'm attempting to implement multiple ways of doing so and am going to compare methods to obtain best results! Thank you.
Please read these two links:
http://www.aishack.in/tutorials/sudoku-grabber-opencv-plot/
How to remove convexity defects in a Sudoku square?
The Problem:
I have an image that I downloaded from google's static map api. I use this image to basically create a "magic wand" type feature where a user clicks. For those interested I am using the graph cut algorithm to find the shape that the user clicked. I then find all the points that represent the border of this shape (borderPoints) using contour tracing.
My Goal:
Straighten out the lines (if possible) and minimize the amount of borderPoints (as much as possible). My current use case are the roofs of houses so in the majority of cases I would hope that I could find the corners and just use those as the borderPoints instead of all the varying points in between. But I am having trouble figuring out how to find those corners because of the bumpy pixel lines.
My Attempts at a Solution:
One simple technique is to loop over the points checking the point before, the current point, and the point after. If the point before and the point after have the same x or the same y then the current point can be removed. This trims the number of points down a little but not as much as I would like.
I also tried looking at the before and after point to see if the current point could be removed if it wasn't within a certain slope range but had little success because occasionally a key corner point was removed because the image was kind of fuzzy and the corner had slightly rounded points.
My Question:
Are there any algorithms for doing this type of thing? If so, what is it (they) called? If not, any thoughts on how to progamatically approach this problem?
This sounds similar to the Ramer–Douglas–Peucker algorithm. You may be able to do better by exploiting the fact that all your points lie on a grid.
Seems to me like you are looking for a polynomial approximation of degree 1.
For a quick answer to your question, you may want to read this: http://en.wikipedia.org/wiki/Simple_regression. The Numerical example section shows you concretely how the equation for your line can be computed.
Polynomial approximations allow you to approach a function, curve, group of point, however you want to call it with a polynomial function of the form an.x^n + ... + a1.x^1 + a0
In your case, you want a line, so you want a function a1.x + a0 where a1 and a0 will be calculated to minimize the error with the set of points you have.
There are various ways of computing your error (called a norm) and minimizing it. You may be interested for example in finding the line that minimizes the distance to any of the points you have (minimizing the max), or in minimizing the distance to the set of points as a whole (minimizing the sum of absolute differences, or the sum of squares of differences, etc.)
In terms of algorithms, you may want to look at Chebyshev approximations and Remez algorithms specifically. All of these solve the approximation of a function with a polynomial of any degree but in your case you will only care about degree 1.
I've been searching far and wide on the seven internets, and have come to no avail. The closest to what I need seems to be The cutting stock problem, only in 2D (which is disappointing since Wikipedia doesn't provide any directions on how to solve that one). Another look-alike problem would be UV unwrapping. There are solutions there, but only those that you get from add-ons on various 3D software.
Cutting the long talk short - what I want is this: given a rectangle of known width and height, I have to find out how many shapes (polygons) of known sizes (which may be rotated at will) may I fit inside that rectangle.
For example, I could choose a T-shaped piece and in the same rectangle I could pack it both in an efficient way, resulting in 4 shapes per rectangle
as well as tiling them based on their bounding boxes, case in which I could only fit 3
But of course, this is only an example... and I don't think it would be much use to solving on this particular case. The only approaches I can think of right now are either like backtracking in their complexity or solve only particular cases of this problem. So... any ideas?
Anybody up for a game of Tetris (a subset of your problem)?
This is known as the packing problem. Without knowing what kind of shapes you are likely to face ahead of time, it can be very difficult if not impossible to come up with an algorithm that will give you the best answer. More than likely unless your polygons are "nice" polygons (circles, squares, equilateral triangles, etc.) you will probably have to settle for a heuristic that gives you the approximate best solution most of the time.
One general heuristic (though far from optimal depending on the shape of the input polygon) would be to simplify the problem by drawing a rectangle around the polygon so that the rectangle would be just big enough to cover the polygon. (As an example in the diagram below we draw a red rectangle around a blue polygon.)
Once we have done this, we can then take that rectangle and try to fit as many of that rectangle into the large rectangle as possible. This simplfies the problem into a rectangle packing problem which is easier to solve and wrap your head around. An example of an algorithm for this is at the following link:
An Effective Recursive Partitioning Approach for the Packing of Identical Rectangles in a Rectangle.
Now obviously this heuristic is not optimal when the polygon in question is not close to being the same shape as a rectangle, but it does give you a minimum baseline to work with especially if you don't have much knowledge of what your polygon will look like (or there is high variance in what the polygon will look like). Using this algorithm, it would fill up a large rectangle like so:
Here is the same image without the intermediate rectangles:
For the case of these T-shaped polygons, the heuristic is not the best it could be (in fact it may be almost a worst case scenario for this proposed approximation), but it would work very well for other types of polygons.
consider what the other answer said by placing the t's into a square, but instead of just leaving it as a square set the shapes up in a list. Then use True and False to fill the nested list as the shape i.e. [[True,True,True],[False,True,False]] for your T shape. Then use a function to place the shapes on the grid. To optimize the results, create a tracker which will pay attention to how many false in a new shape overlap with trues that are already on the grid from previous shapes. The function will place the shape in the place with the most overlaps. There will have to be modifications to create higher and higher optimizations, but that is the general premise which you are looking for.
I have problem of packing 2 arbitrary polygons. I.e. we have 2 arbitrary polygons. We are to find such placement of this polygons (we could make rotations and movements), when rectangle, which circumscribes this polygons has minimal area.
I know, that this is a NP-complete problem. I want to choose an efficient algorithm for solving this problem. I' looking for No-Fit-Polygon approach. But I could't find anywhere the simple and clear algorithm for finding the NFP of two arbitrary polygons.
The parameter space does not seem too big and testing it is not too bad either. If you fix one polygon, the other ploygon can be shifted along x-axis by X, and shifted along y-axis by Y and rotated by r.
The interesting region for X and Y can be determined by finding some bounding box for for the polygons. r of course is between and 360 degrees.
So how about you tried a set of a set of equally spaced intervals in the interesting range for X,Y and r. Perhaps, once you found the interesting points in these dimensions, you can do more finer grained search.
If its NP-complete then you need heuristics, not algorithms. I'd try putting each possible pair of sides together and then sliding one against the other to minimise area, constrained by possible overlap if they are concave of course.
There is an implementation of a robust and comprehensive no-fit polygon generation in a C++ library using an orbiting approach: https://github.com/kallaballa/libnfporb
(I am the author of libnfporb)