Manual Image Processing II

This might be one simple work. 907-pairs images (will be mailed later) have to be 'digitized' using the following procedures within 30 days. Computer knowledge not required - great for someone who works at home. [login to view URL] 0. Uncompress, and run program 1. Open the [login to view URL] program, you can add the two images using the two 'Load Image' buttons (form work folder), the loaded imaged are shown in the two panels ([login to view URL] in left, [login to view URL] in right). 2. Check the stereo window on the pictures, choose the two convergence points (note below) of the pair. Move the mouse over the image, a magnify glass with radius 50 pixels will appear. In the magnify glass, the image around the point is shown in their original size. 3. Choose: Point 1 (above calculate button) Place mouse cross-bar (in magnify glass) to 'point 1' of left image, left-click Place mouse cross-bar to 'point 1' of right image, click 4. Choose: Point 2 Place mouse cross-bar to 'point 2' of left image, click Place mouse cross-bar to 'point 2' of right image, click The coordinates of the points are shown in the top two panels. Remember to select point 2 in the drop down list when you are done with point 1. 5. Click Calculate button when all the four points are selected. The results are shown in the 'Transformation Parameters' Panel. 6. Click Transformation button, a dialog is shown with transformation progress. It may take several seconds to complete if the images are large. Since the left and right image are not exactly the same depth, the images may not be aligned exactly especially for those points far away from the four mapping points. 7. The transformation parameters and image names are also saved into '[login to view URL]' text file. such as '[login to view URL] [login to view URL] -88.2 13.333 -0.0003'. This file will be appended for further batch automatic process. 8. E-mail me this '[login to view URL]' file. --------------------------------------------------------------------- Note: Determine the convergence point of the pair from step1 folder, this often involves looking at the image in 3D and asking yourself which point do you want to be at the stereo "window" (the spot along the z axis right exactly at the screen). Very often it will be the foremost spot in the image, but if you want something to come through the "window" (out of the screen) you will want to converge on something behind the near-object. The convergence point of a scene has ZERO parallax - it is in the exact same spot in each image of the stereo pair.