Bon Courage :)

Problem 1: Cleaning image cell.pgm

INPUT: cell.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT (1C)
	The small white spots and black holes in cell.pgm correspond to noise due to the acquisition process. They do not carry any information about the cells under study. Such noise is called (binary) salt and pepper noise or speckles. Problem 1 is about removing salt and pepper noise from this image.

Guideline.

• 1A - Remove the white spots (the smallest) visible in the black background.
• 1B - Remove the black holes within the white cells/blobs.
  
• 1C - Remove salt and pepper noise: produce an image like the one above on the right.
•  1D,E,F - Same goal as 1A,B,C but without "moving" the contours of the cells (to be done after step 2B).

Tools.
You could use (compositions of) the dilation and erosion programs that you have implemented during the first practical session of the course. Instead, to save time, you can use the programs of Pink Library, which are made available to you, through Linux terminals of the provided Working Environment:

• 'esedit.tcl'
  
• 'dilation'
• 'erosion'
• 'dilatball'
• 'erosball'

In order to visualize images you can use from a terminal the command 'display' followed by the name of the images to visualize.
 
Additional Tools for solving two stars questions are shown at the end of the page.

Problem 2: Filling the holes of cell3.pgm 

INPUT: cell3.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT (2B)
	Image cell3.pgm contains cells (in white). The absence/presence/number of holes allows different kinds of cells to be distinguished. A first step towards identification of the different cells consists of extracting the holes. This can be done by i) filling all holes; and ii) taking the difference between the initial image and the one  without holes. Problem 2 is about filling the holes of cell3.pgm

Guideline.

• 2A - Fill the black holes in the interior of the white blobs or cells.
• 2B - Fill the holes while preserving the shape of the external contours of the cells exactly as in the original image (see for instance the expected result shown above). To reach this goal, use operaor geodilat.

Problem 3: Extracting descending objects of lines.pgm

INPUT: lines.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT (2B)
	A common situation in image analysis consists of extracting objects that spread the image following one (or several) preferred direction(s). For instance, the sky line in a natural image is in general horizontal.The image lines.pgm is a synthetic one produced to illustrate how one can extract such kind of objects. Problem 3 is about extracting "descending objects" (from top left to bottom right) of image lines.pgm.

Guideline.

• 3A - Extract Lines which are descending: produce an image like the one above on the right.

Tools.
You can use any tool presented for the previous problems.

Problem 4: Smoothing the aliasing in jagged.pgm 

INPUT: jagged.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT (4B)
	Have a look at this link. More details are provided here.

Guideline.

• 4A - Smooth the black circle of the image jagged.pgm in order to obtain regular contours.
• 4B - Same as 4A, but the resulting black circle must be neither larger than the one of jagged.pgm nor smaller.

Tools.
You can use any tool presented for the previous problems.

The three star item 4B can be solved by a combination of tools presented until now. It can also be solved by a tool presented in the 'Additional Tools' shown at the end of the page.

Problem 5: Extracting the contours of cell2.pgm 

INPUT: cell2.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT
	The contour of a shape is useful for analyzing its geometry. For instance a shape (like a cell in image cell2.pgm) is circular when the ratio between the square of its perimeter and its area tends to  4, whereas it is elongated when this ratio is high. Problem 5 is about contour extraction of cell2.pgm.

Guideline.

• 5A - Extract the contours of cell2.pgm: produce an image like the one above on the right.

Tools.
You can use any tool presented for the previous problems. Furthermore you can use:

• 'min' to compute an union.
• 'max' to compute an intersection.
• 'sub' to compute a set difference: 

Problem 6: Cleaning image sofitel.pgm

INPUT : sofitel.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT (6B)
	The small bright and dark spots in sofitel.pgm correspond to noise that may be due to the acquisition or transmission process for instance. They do not carry any information about the picture and disturb visualization. Such noise is called (grayscale) salt and pepper noise or speckles. Problem 6 is about removing salt and pepper noise from this image.

Guideline.

• 6A - Remove salt and pepper noise. At this step, loss of contrast and distortion of the contours are accepted.
• 6B - Remove salt and pepper noise: produce an image like the one above. At this step loss of contrast or distortion of the contours are no more allowed.

Tools.
To obtain topographic surfaces from a grayscale images save the program '2d2relief.tcl' (dowload it here) in your working directory and enter './2d2relief.tcl'in a terminal to obtain a short help. You can visualize the reliefs with the command 'multirender' followed by the names of the reliefs.

Note that the programs 'dilation', 'erosion', 'opening', and 'closing' accept grayscale images whereas 'erosball', 'dilatball', 'openball' and 'closeball' do not.

Additional Tools for solving two and three stars questions are shown at the end of the page.

INPUT: sofitel2.pgm	DESCRIPTION OF THE PROBLEM	EXPECTED RESULT
	In order to detect the contours of the object in image sofitel2.pgm, it is useful to start by "sketching its edge": computing an image whose values are high near the contours. Problem 7 is about producing such an image.

Guideline.

• 7A - Sketch the edges: produce an image like the one above on the right.

Tools.
You can use any tool presented for the previous problems.

2. Alternate sequential filter

•  'asf'