Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 7 views

    TD Active

    Uploaded by

    soukaina.chrit
    The document describes using an active contour or "snake" to approximate the contour of an object in a binary image. The snake is defined by three points that minimize an energy function consisting of internal and external energies. An external energy function is proposed that returns small values on edges and high values on flat regions. The algorithm iteratively moves each snake point to its neighbor with minimum energy until convergence. Examples are given initializing the snake points and showing their successive positions until convergence, with comments on why one result is not appropriate. An expression for internal energy is requested that favors short snake lengths and ensures the snake remains triangular in shape.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    TD Active

    Uploaded by

    soukaina.chrit
    7 views2 pages
    The document describes using an active contour or "snake" to approximate the contour of an object in a binary image. The snake is defined by three points that minimize an energy function consisting of internal and external energies. An external energy function is proposed that returns small values on edges and high values on flat regions. The algorithm iteratively moves each snake point to its neighbor with minimum energy until convergence. Examples are given initializing the snake points and showing their successive positions until convergence, with comments on why one result is not appropriate. An expression for internal energy is requested that favors short snake lengths and ensures the snake remains triangular in shape.

    Original Title

    td-active

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document describes using an active contour or "snake" to approximate the contour of an object in a binary image. The snake is defined by three points that minimize an energy function consisting of internal and external energies. An external energy function is proposed that returns small values on edges and high values on flat regions. The algorithm iteratively moves each snake point to its neighbor with minimum energy until convergence. Examples are given initializing the snake points and showing their successive positions until convergence, with comments on why one result is not appropriate. An expression for internal energy is requested that favors short snake lengths and ensures the snake remains triangular in shape.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    7 views2 pages

    TD Active

    Uploaded by

    soukaina.chrit
    The document describes using an active contour or "snake" to approximate the contour of an object in a binary image. The snake is defined by three points that minimize an energy function consisting of internal and external energies. An external energy function is proposed that returns small values on edges and high values on flat regions. The algorithm iteratively moves each snake point to its neighbor with minimum energy until convergence. Examples are given initializing the snake points and showing their successive positions until convergence, with comments on why one result is not appropriate. An expression for internal energy is requested that favors short snake lengths and ensures the snake remains triangular in shape.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Université de Rouen Normandie, INSA de Rouen

    M2 Data Science - Science des données

    Advanced image processing

    Exercice on active contours

    We want to approximate the contour of an object in an image by an active contour. We know that
    the shape of the object to detect is close to a triangle. Hence we propose to detect the contour of
    this object by an active contour, or a "snake" defined by three points x1, x2 and x3. The best
    position of the active contour is the one that minimizes an energy function E which is the sum of
    an internal energy Eint and an external energy Eext defined by :

    E = Eint + Eext

    For the sake of simplicity, we assume that the image we want to analyze with the active contour
    has already been pre-processed and is now binary. By convention, the pixels belonging to the
    object to be detected are at 0 and the pixels belonging to the background are set to 255. Figure 1-a
    gives an example of such an image and figure 1-b the approximation obtained with the snake on
    this image.

    Fig 1-a: Binary image to be segmented Fig 1-b: Segmentation result with a triangle active contour

    1) Propose an external energy function that will allow the detection of the object contours by
    such a method, ie that will return small values on edges and high values on flat region.

    The values of the external energy calculated on a 10x10 binary image is now to defined in the
    matrix below:

    1
    We apply the following algorithm:

    1-> Position an initial active contour (which is a triangle) ;


    2-> While the global energy of the active contour continues to decrease do :
    -> For each point of the active contour do :
    -> Search for neighboring pixels of minimum energy (taking into account a 8
    neighboring pixels);
    -> Move the active contour point to its new position.

    The following rules apply:


    - the progression step of the points of the active contour is 1 pixel at each iteration;
    - if several neighboring pixels have the same minimum energy at step 2 of the algorithm,
    one of the pixels is randomly selected;
    - if some pixels in the point the neighborhood are defined outside of the image, these
    particular pixels are not considered.

    2) Give the positions of the successive active contour points until convergence with the
    following initialization: x1 = (6,4), x2 = (5,5), x3 = (6,6).
    In addition, draw the points on the figure above.

    NB: the coordinates of a point correspond to the indices as follows: (6,4) => i=6, j=4
    NB2: the internal energy is supposed to be constant.

    3) Same question as 2) but with the following initialization:


    x1 = (10,2), x2 = (2,5), x3 = (10,6).

    4) Why is the result not appropriate?

    5) Propose an arithmetic expression for the calculation of the internal energy as a function of
    the points x1, x2 and x3 so that this energy is no longer a constant but a function
    expressing the fact that:
    a) the length of the resulting contour is as short as possible;
    b) the final contour is a triangle (two points of the active contour cannot converge to the
    same point).

    You might also like