Image Processing and Neural network

hi,I'm a researcher regarding neural network and image processing on matlab. its all about ASL to text.. by using bare hands ASL and a black background the signs will be converted in to TEXT by comparing the shot image to the database.. well, I'm having a difficulty regarding the comparison.. can someone help me.. also with neural network..


with so much gratitude,

kaigi

Comments

  • : hi,I'm a researcher regarding neural network and image processing on
    : matlab. its all about ASL to text.. by using bare hands ASL and a
    : black background the signs will be converted in to TEXT by comparing
    : the shot image to the database.. well, I'm having a difficulty
    : regarding the comparison.. can someone help me.. also with neural
    : network..

    I suggest the following basic approach:

    1. Capture the relevant portion of the screen
    2. Isolate individual glyphs (characters)
    3. Pre-process the characters (center, improve gain and contrast, etc.)
    4. Extract informative features
    5. Recognize individual characters based on features
    6. Celebrate!

    In step 4, assuming that a standard-sized region is being used (n x m pixel rectangle), then some features to consider might be: horizontal and vertical projection profiles (row and column sums) and the total count of "foreground" pixels. Feature extraction can greatly reduce the size of the data, and emphasize relevant information at the same time.

    Step 5 will be performed by a neural network, linear discriminant, nearest neighbor, etc.


    -Will Dwinnell
    [link=http://matlabdatamining.blogspot.com/]Data Mining in MATLAB[/link]



Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Categories

In this Discussion