| Best methods for finding a threshold edge, like a coastline? [message #85433] |
Sat, 03 August 2013 11:13 |
kagoldberg
Messages: 26
Registered: November 2012
|
Junior Member |
|
|
In a 2D image, I'm looking for the best way (accurate, fast) to find the contours of an edge, defined by a threshold along a rough slope. I've come up with three ways to do it, but they give different results.
Imagine a 2D image with a slope in the x direction, and some roughness, like a north-south coastline. Define a threshold value, and use interpolation to find the contours of the edge at every vertical position. Ideally, I'd like to have a continuous value in the x direction, and evenly spaced values in y (i.e. one x position per row).
**Method 1, use INTERPOL on every line. Like this.
for i=0,Ny-1 do $
edge[i] = interpol(x, img[*,i], threshold)
**Method 2, REBIN the array, stretching in x and then binarize the thresholding (1=above, 0=below) for quick counting. Like this. The result is still discrete, but the steps are as fine as you want to achieve with the stretching.
img1 = rebin(img, 1000, N) GT threshold ;--- stretch
Edge = total(img1, 1)/stretch_factor ;--- this is a 1D array that counts the points above threshold
**Method 3, use the CONTOUR procedure to calculate the 'path' of the edge.
contour, img, levels=threshold, /PATH_DATA_COORDS, PATH_XY=xy, CLOSED=0
;--- now xy is a 2D array, and xy[0,*] contains the edge positions, one per line.
Discussion
Method 1 is a for-loop solution, so we know it's not optimized.
Method 2 can be fast, but we have to choose the degree of precision/time.
Method 3 is lightning fast (4-6x faster!) but the results are inconsistent with 1 and 2. Could this be because contour is applying a non-linear interpolation?
Does anyone have any insight to share on this?
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
