Basic concepts

Table of Contents

• Generalities
• Data objects
  • Images
  • Structuring Elements
  • Blobs
  • Points
  • Graphs
• Functions and Methods
• GUI - Graphical User Interface
• I/O - Reading and Writing images

Generalities

Data objects

Images

Images are the basic object type in Smil. Smil was created to manipulate images using mathematical morphology techniques.

Smil Images can basically be declared in one of the following data types :

• Gray level images : UINT8, UINT16 and UINT32 - meaning unsigned integer of size 8, 16 or 32 bits;
• Binary images : are defined in one of the data types used for gray images but having only two possible values (usually 0 and the biggest value for that data type); in one of these kinds but having only two possible values (usually 0 and the biggest value for that data type)
• Colour images : support to these kind of images is limited. They're represented in their three channels. These images can be converted into gray images or each channel can be considered as a gray image and handled this way.

Smil images can be of type 1D, 2D or 3D.

From the computer point of view, an image is an instance of a class with two parts :

• metadata : side information needed to identify and handle image data : size, image type, name, ... You can't access this values directly but you can read or modify some values, thanks to Image class methods.

• pixel values : organized as an array. Pixel values can be accessed individually using class methods or by some pointer directly to the array.

  When called from Python (module smilPython), Smil images can be handled by Numpy functions or converted in both directions.

Structuring Elements

Structuring Elements ar just a small set of points used to probe the image under study.

There are some pre defined Structuring Elements :

• 2D
  • HexSE(UINT size) - hexagonal grid
  • VertSE(UINT size)
  • HorizSE(UINT size)
  • CrossSE(UINT size)
  • SquSE(UINT size)
  • LineSE(int length, int theta) - lines with arbitrary length and angle starting from center
• 3D
  • Cross3DSE(UINT size)
  • CubeSE(UINT size)
  • RhombicuboctaedronSE(UINT size)
  • Line3DSE(int length, int theta, int zeta) - 3D line with arbitrary lenght and angle starting from center

Blobs

Blobs are just non connected regions in images. A set of pixels sharing a common property in binary or gray images.

One can use a blob to evaluate parameters of the region : area (count of pixels), volume, statistics (mean, standard deviation, ...), moments, barycenter, inertia matrix, ...

As a data type, a blob is a C++ map, or Python dict having the label (region identifier) as a key and a transparent content with all data needed to make calculations on the region.

Blobs can be used directly to get information about regions in a binary image as below :

 import smilPython as sp

 # read input image
 imIn = sp.Image("balls.png")

 # label it
 imLabel = sp.Image(imIn, 'UINT16')
 sp.label(imIn, imLabel)

 # compute labels
 blobs = sp.computeBlobs(imLabel)

 # get some values on each labelled region
 areas = sp.blobsArea(blobs)
 bboxs = sp.blobsBoundBox(blobs)

Or you can use blobs as a mask to get informations about regions in a gray level image. To do that, you should, before all, segment it and get a binary mask.

 import smilPython as sp

 # read input image
 iName = "https://smil.cmm.minesparis.psl.eu/images/balls.png"
 imIn = sp.Image(iName)

 # segment it to create a mask and label binary image
 imThr = sp.Image(iIn)
 sp.topHat(imIn, imThr, sp.hSE(20))
 sp.threshold(imThr, imThr)
 imLabel = sp.Image(imIn, "UINT16")

 # compute blobs from labeled image
 blobs = sp.computeBlobs(imLabel)

 # get some values from mask
 areas = sp.blobsArea(blobs)
 bboxs = sp.blobsBoundBox(blobs)

 # get some values on each labelled region
 barys = sp.blobsBarycenter(imIn, blobs)
 # mean and standard deviation of pixel values
 means = sp.blobsMeanVal(imIn, blobs)
 # first and second moments
 moments = sp.blobsMoments(imIn, blobs, True)
 # inertia matrix
 inertia = sp.blobsInertiaMatrix(imIn, blobs, True)

Some functions available : blobsArea(), blobsVolume(), blobsMoments(),blobsMinVal(), blobsMaxVal(), blobsRangeVal(), blobsValueList(), blobsModeVal(), blobsMeanVal(), blobsBarycenter(), blobsBoundBox(), blobsEntropy() and drawBlobs().

Points

Graphs

Functions and Methods

GUI - Graphical User Interface

I/O - Reading and Writing images