Digital image processing using MATLAB using three methods

Question

I have homework in MATLAB. I have to use 3 image processing methods. Therefore, I have to complete the task and then solve it using 3 methods (e.g. threshold value, segmentation, morphology, restoration, histogram alignment, noise removal ...). I need an idea and how to solve it, will you help me? :)

Thank.

• In edition:

I found this in some book ... Do you have an idea? Is it possible to restore image a for image i ?

Note . Some solution is listed below. But to be honest, I didn’t understand :( Can you explain this to me?

Answer 1

Restoring this picture is very difficult ... Therefore, I decided to change the task.

Task and solution are discussed here:

http://geogeeks.net/2011/03/18/digital-image-processing-using-matlab/

Answer 2

You can, for example, try to isolate an object in three different ways.

Do it in Mathematica. (MATLAB is homework).

Call our image i :

i =

And try isolating the mask called mask :

mask =

See code examples:

(* First Method, by Image Correlation*)
x = ImageCorrelate[ i, mask, EuclideanDistance];
r = Position[ImageData@Binarize[x, 0.2], 0, Infinity];
(*Show that we found the right spot *)
ImageCompose[i, 
 ColorNegate@
  mask, {0, Dimensions[ImageData[i]][[1]]} - {-1, 1} Reverse[r[[1]]]]

Result:

(* Second method, separating channels, 
   thresholding and deleting small components*)

r = DeleteSmallComponents@Binarize[#, .99] &@
   ColorNegate[ColorSeparate[i][[3]]];
ImageMultiply[i, r]

Result:

(* Third method, extracting the exact color *)
Image[ImageData[i] /. {1., 0.6, 0.} -> {a} /. {_, _, _} -> {0, 0,0} /. 
                                       {a} -> {1., 0.6, 0.}]  

Result:

NTN!

Answer 3

, . , . .

, :

TotalVariationFilter[image, 1, Method -> "Laplacian"]  

And then you must deconvolute the blur of diagonal movement. You need a kernel like this:

That, when applied to de-noise, it turns out:

ImageDeconvolve[denoisedImage, kernel, Method -> "RichardsonLucy", 
 MaxIterations -> 15]

The image is not perfect, but I hope this gives you an idea of ​​what can be done.