Image processing
in [Matlab]
|
From: Harsha Dissanayake on 22 Apr 2010 12:32 I have a problem in image processing. I want to know how do I calculate a height by using a pixel value in a color image. In here my approach is to measure height of a teak tree. I am using color images of a teak trees.
From: Kadi on 22 Apr 2010 12:54 In order to convert pixel to inches you need to know how many pixels are there in an inch (pixels per inch-ppi) and this again depends on the height and width of your image. Kadambari "Harsha Dissanayake" <lantrapop(a)gmail.com> wrote in message <hqptmp$4j5$1(a)fred.mathworks.com>... > I have a problem in image processing. I want to know how do I calculate a height by using a pixel value in a color image. In here my approach is to measure height of a teak tree. I am using color images of a teak trees.
From: Sean on 22 Apr 2010 12:55 "Harsha Dissanayake" <lantrapop(a)gmail.com> wrote in message <hqptmp$4j5$1(a)fred.mathworks.com>... > I have a problem in image processing. I want to know how do I calculate a height by using a pixel value in a color image. In here my approach is to measure height of a teak tree. I am using color images of a teak trees. Can you post your image on a website somewhere so we can see it?
From: Walter Roberson on 22 Apr 2010 13:13 Harsha Dissanayake wrote: > I have a problem in image processing. I want to know how do I calculate > a height by using a pixel value in a color image. In here my approach is > to measure height of a teak tree. I am using color images of a teak > trees. You need a reference height to calibrate against, preferably as close to the target section as possible (to prevent parallax errors) but not so far away that the reference height markings are distorted. Reference heights can be either explicit (e.g., an object of known height placed in view) or implicit (e.g., the size of human eyes varies little as they do not grow over time, though the distance between human eyes may vary.) As well as a reference height you need a reference distance to the target section, or at least something by which you can calculate the distance triganometrically. Taking perspective into account is important, and depending on the quality of the camera and the distance to the target section, it may also be important to correct for lens distortions of various kinds -- e.g., a "wide-angle lens" does not provide constant distances across the image. And with something like a tree, there can be difficulty in deciding what exactly the "height" of the tree is. Do you include the leaves? When the leaves might in fact be quite big? When the leaves might change angle with different times of day? For something like a Banana "tree", do you include the pseudo-stems that the fruit (technically berries) grow upon? Sometimes tree heights are defined in terms of the highest horizontal branches -- which can end up discarding the effect of the soft "crown" on something like a pine tree (which does tend to grow horizontal branches). Sometimes tree heights are defined in terms of the highest "hardened" (second year) growth, knowing that the tree may well send out extensive "suckers" that will mostly die off and fall off (e.g., a willow tree.) So... you have two important tasks to start here: come up with a calibration system, and come up with a robust (and deducible from the image) definition of what it means for the trees to be particular heights. You have an additional difficulty that teak often grows in or adjacent to swamps, and the edge of swamps are often of fairly variable geography; if a teak tree grows on a hump in the middle of a swamp, then to be fair, you have to somehow subtract off the height of the part of the hump the tree is growing on in order to accurately calculate the height of the tree... though if the tree is growing _in_ the swamp then you have to know how deep the swamp is there to calculate the tree height... which in turn gets you into arguments about what depth in the much or grasses in a swamp you should consider to be base point for height calculation purposes. There simply isn't an easy answer to your question, especially for teak (but mangrove would be probably even worse, since you would then have to deal with questions about where one mangrove tree ends and another begins.) You have some decisions to make.
|
Pages: 1 Prev: Spring 2010 MATLAB Contest Schedule Next: How to generate multiple simulations in SIMULINK |