MI Pre 1974 UK Counties Outlines
Does anyone know where I could find pre-1974 UK county outlines in either MapInfo or AutoCAD readable form? I don't mind having to do some clean-up to get them converted to regions. This is for presentation of density and movement data in a one-name study so I'm hoping to minimize costs. Thanks, John -- To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put "unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
Re: MI Looking for Nearest Object Distance Functions
Stephen: I had a similar problem - I needed to find the distance from a number of given points to their nearest contour line of a given height. This was before I found this list, so I solved it in a typical O'Brien brute-force manner, using a kind of binary chop algorithm: For each point I created a buffer (with 32 sides as a compromise between size and circularity) and checked for intersection with the line. If no intersection, I doubled the buffer radius and tried again. Once intersection had been achieved I used the usual binary chop method to successively change the radius by halving increments (adding if no intersection, subtracting if intersection) until the increment size was within the tolerable error (I used 0.1 metre). It worked, but is NOT elegant, and I'll send a .mb file should you so wish. But I DO hope that someone has a better answer - when applied to a grid of 2,000,000 points it took several weeks of overnight runs (on a P166) to complete! Would I do it that way again? No, I'd post on this list and hope! - Original Message - From: Stephen R. Riese <[EMAIL PROTECTED]> To: MapInfo <[EMAIL PROTECTED]> Sent: Friday, December 31, 1999 9:01 PM Subject: MI Looking for Nearest Object Distance Functions > > Greetings, > > I'm new to MapInfo/MapBasic and this list server. I'm looking for a > function that will return the distance from a given point to the nearest > object of a certain type. For example, from point (x, y) to the nearest > house (point object), road or river (linear object) or nearest forrest > (area object). All of the functions I've come across give the distance > between two known points. I need to find the closest object first (e.g., > which house or river is closest -- hard to do) and then determine the > distance (easy to do). > > Is there some way to avoid an exhaustive search? I need to make tens of > thousands of these calculations, each one selecting the nearest of up to > hundreds of similar objects. In addition, each linear and area object are > defined by two or more points. I assume I'll need to determine which point > within each of these objects is closest to the known (x, y) point. > > Thanks for any help you can provide and Happy New Year! > > [EMAIL PROTECTED] > -- > To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put > "unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED] -- To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put "unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
Re: MI Looking for Nearest Object Distance Functions
"a simple Print (to a window) statement in each cycle brings the search down to a crawl." Too right! I put lots of prints in each run at first until I realised how much time it took - Original Message - From: Stephen R. Riese <[EMAIL PROTECTED]> To: O'Brien, Graham [MI] <[EMAIL PROTECTED]> Sent: Saturday, January 01, 2000 4:18 PM Subject: Re: MI Looking for Nearest Object Distance Functions Message text written by "Graham O'Brien" >Stephen: I had a similar problem - I needed to find the distance from a number of given points to their nearest contour line of a given height. This was before I found this list, so I solved it in a typical O'Brien brute-force manner, using a kind of binary chop algorithm: For each point I created a buffer (with 32 sides as a compromise between size and circularity) and checked for intersection with the line. If no intersection, I doubled the buffer radius and tried again. Once intersection had been achieved I used the usual binary chop method to successively change the radius by halving increments (adding if no intersection, subtracting if intersection) until the increment size was within the tolerable error (I used 0.1 metre). It worked, but is NOT elegant, and I'll send a .mb file should you so wish. But I DO hope that someone has a better answer - when applied to a grid of 2,000,000 points it took several weeks of overnight runs (on a P166) to complete! Would I do it that way again? No, I'd post on this list and hope!< Graham, Thanks. Your method gives me a few ideas I hadn't considered yet. Actually, I had thought about incrementally expanding a radius of search to find objects, but had no clue as to how to go about it. The expanding buffer intersection algorithm (EBIA??) seems reasonable. I like the idea of bracketing the object separation and halving the distance until within tolerance. Kind of like the typical strategy to minimize the number of guesses to pick a hidden number. I'm experimenting now with a small grid that has 750 cells and must find the closest of about 50 point objects. It actually runs faster than I expected (P200) using the brute-force method. The time is measured in minutes, not tens of minutes, which is a good sign. I also learned how slow graphic operations are compared to numeric operations. Inserting a simple Print (to a window) statement in each cycle brings the search down to a crawl. My eventual grid will likely have tens of thousands of cells and need to find (for each cell) the closest of hundreds or low thousands of objects in each of ten to fifteen categories. The old combinatorial monster is creeping up my back -- much like your problem with 2 million cells. Thanks again and Happy New Year! Steve [EMAIL PROTECTED] -- To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put "unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
