Summary of Weekly Meeting - (10)

I. Trying 2 possible solutions:

    A. Solution 1: improve the way to maintain the relationship of homography
       
         Averaging the error from each set of 3 points which is from 3 adjacent color markers
             -> Generating Transformation matrix from each cases (...totally 4 cases in 4 color markers)
             -> Computing the mapping results from Hand-held camera frame to Kinect frame with a viewpoint
             -> Averaging the 4 mapping results from each transformation matrix
             -> Verifiying...

    B. Solution 2: improve the way to locate the central point
       
         Using contour points of each color marker instead of just doing color thresholding
             -> Eclipse is better, which have a function can compensate the misleading contour points
             -> After locating contour points for each color marker, finding the position of central points
             -> Taking these central point as fixed points
             -> Applying 'Solution 1' during the process of verifying


          * The issue of using rectangle:
              1. The shape of rectangle may alter when tilting as showing below
              2. While the shape changes from rectangle into a trapezoid, some contour points on the short side
                  may miss because of the decreasing of resolution / image quality / shaping
              3. Eclipse has a friendly function can handle the misleading / missing contour points, which is easier
                  than coping with  a rectangle

    C. Solution 3: any ways after trying the former solution

II. Finishing the modification of code based on Solution 1: 
     1. Not seeing to much improvement
     2. Next, verifying the influence caused by the position of fixed points
     3. Then, change the color marker from rectangle ones to circle ones (... plus contour tracking)