0 Replies - 450 Views - Last Post: 20 November 2017 - 12:36 PM Rate Topic: ***-- 2 Votes

#1 KhaosFury  Icon User is offline

  • New D.I.C Head

Reputation: 0
  • View blog
  • Posts: 19
  • Joined: 24-November 14

Image Analysis

Posted 20 November 2017 - 12:36 PM

Hello!

I have an interesting problem. I've got an application which generates a sort of Heightmap based on an image as such:
https://imgur.com/a/hRcvc

I have been trying, for awhile now, to think of a way I could create an algorithm to analyze the shape of the anything above a certain Z-Axis level.
My ultimate goal is to determine how irregular the shape is by comparing the image to a rectangle and returning an integer (From 0-1) based on how similar the white object is to a rectangular box. Or I suppose, from the algorithm's point of view, how many straight edges are present in the image.

Currently, I generate the heightmap to get X, Y, and Z as such:
Private zData(,) As Single
                For y As Integer = 0 To bmp.Height - 1
                    For x As Integer = 0 To bmp.Width - 1
                        zData(x, y) = bmp.GetPixel(x, y).GetBrightness
                    Next
                Next



And for Z:
		
		For y As Integer = 0 To data.GetUpperBound(1)
			For x As Integer = 0 To data.GetUpperBound(0)
				minZ = Math.Min(zData(x, y), minZ)
				maxZ = Math.Max(zData(x, y), maxZ)
			Next
		Next



So far, I'm thinking I'd linearly parse the entire image (or array of values) along the Y-Axis, and determine how many points along the Y-Axis contain a value of Z >= 0.5 (or whatever value I am wanting to find). However many hits it gives me in a row will contribute to how linear the image is along that axis, and then again along the X axis?
Problem with that idea is 2 things:
I can't wrap my head around if that would be accurate enough
I am not sure how I can write a loop which can do this

Any ideas, or guidance would be extremely appreciated!

Is This A Good Question/Topic? 0
  • +

Page 1 of 1