inComplete42

05computerVision.ipynb

@@ -222808,7 +222808,7 @@
     "fig.update_scenes(camera_projection_type='orthographic')\n",
     "\n",
     "fig.show()\n",
-    "fig.write_html('./data/stereo/stereo-b.html')"
+    "fig.write_html('./data/stereo/stereo.html')"
    ]
   },
   {
@@ -223446,9 +223446,9 @@
    "source": [
     "- **Consider what we have just done.**\n",
     "\n",
-    "        **a. Calculated a reprojection error** _(backprojecting 3D coordinates onto the 2D image to compare with the original keypoints)_;  \n",
-    "        **b. Solved for the orientation of the cameras with a RANSAC-based P-$n$-P** _(when we already have orientation parameters)_ and  \n",
-    "        **c. Optimised the solution with a Bundle Adjustment** _(here we call this the Minimised error)_\n",
+    "        **- Calculate a reprojection error** _(2.c.i. - backprojecting 3D coordinates onto the 2D image to compare with the original keypoints)_;  \n",
+    "        **- Solve for the orientation of the cameras with a RANSAC-based P-$n$-P** _(2.d. - when we already have orientation parameters from 2.b.)_ and  \n",
+    "        **- Optimise the solution with a Bundle Adjustment** _(2. e and f. - here we call this the Minimised error)_\n",
     "\n",
     "**Why go through all these refinements? What purpose does it serve to repeatedly search for the solution that minimises the error between observed** (2D image) **and predicted** (3D world) **points AND continouosly determine the most effective viewing parameters** _(i.e., camera pose and possibly intrinsic calibration and radial distortion)_**?**"
    ]