You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
On the diagram, red rectangle are firmware settings that are configurable via the API. Gray rectangles are settings that that are not yet
105
+
exposed to the API. We plan on exposing as much configurability as possible, but please inform us if you would like to see these settings
106
+
configurable sooner.
107
+
108
+
If you click on the image, you will be redirected to the webapp. Some blocks have notes that provide additional technical information.
109
+
110
+
Currently configurable blocks
111
+
*****************************
79
112
80
-
Left-Right Check
81
-
****************
113
+
.. tabs::
114
+
115
+
.. tab:: Stereo Mode
116
+
117
+
.. tabs::
118
+
119
+
.. tab:: Left-Right Check
120
+
121
+
Left-Right Check or LR-Check is used to remove incorrectly calculated disparity pixels due to occlusions at object borders (Left and Right camera views
122
+
are slightly different).
123
+
124
+
#. Computes disparity by matching in R->L direction
125
+
#. Computes disparity by matching in L->R direction
126
+
#. Combines results from 1 and 2, running on Shave: each pixel d = disparity_LR(x,y) is compared with disparity_RL(x-d,y). If the difference is above a threshold, the pixel at (x,y) in the final disparity map is invalidated.
127
+
128
+
.. tab:: Extended Disparity
129
+
130
+
The :code:`extended disparity` allows detecting closer distance objects for the given baseline. This increases the maximum disparity search from 96 to 191.
131
+
So this cuts the minimum perceivable distance in half, given that the minimum distance is now :code:`focal_length * base_line_dist / 190` instead
132
+
of :code:`focal_length * base_line_dist / 95`.
82
133
83
-
Left-Right Check or LR-Check is used to remove incorrectly calculated disparity pixels due to occlusions at object borders (Left and Right camera views
84
-
are slightly different).
134
+
#. Computes disparity on the original size images (e.g. 1280x720)
135
+
#. Computes disparity on 2x downscaled images (e.g. 640x360)
136
+
#. Combines the two level disparities on Shave, effectively covering a total disparity range of 191 pixels (in relation to the original resolution).
85
137
86
-
#. Computes disparity by matching in R->L direction
87
-
#. Computes disparity by matching in L->R direction
88
-
#. Combines results from 1 and 2, running on Shave: each pixel d = disparity_LR(x,y) is compared with disparity_RL(x-d,y). If the difference is above a threshold, the pixel at (x,y) in the final disparity map is invalidated.
138
+
.. tab:: Subpixel Disparity
89
139
90
-
Extended Disparity
91
-
******************
140
+
Subpixel improves the precision and is especially useful for long range measurements. It also helps for better estimating surface normals
92
141
93
-
The :code:`extended disparity` allows detecting closer distance objects for the given baseline. This increases the maximum disparity search from 96 to 191.
94
-
So this cuts the minimum perceivable distance in half, given that the minimum distance is now :code:`focal_length * base_line_dist / 190` instead
95
-
of :code:`focal_length * base_line_dist / 95`.
142
+
Besides the integer disparity output, the Stereo engine is programmed to dump to memory the cost volume, that is 96 levels (disparities) per pixel,
143
+
then software interpolation is done on Shave, resulting a final disparity with 5 fractional bits, resulting in significantly more granular depth
144
+
steps (32 additional steps between the integer-pixel depth steps), and also theoretically, longer-distance depth viewing - as the maximum depth
145
+
is no longer limited by a feature being a full integer pixel-step apart, but rather 1/32 of a pixel.
96
146
97
-
#. Computes disparity on the original size images (e.g. 1280x720)
98
-
#. Computes disparity on 2x downscaled images (e.g. 640x360)
99
-
#. Combines the two level disparities on Shave, effectively covering a total disparity range of 191 pixels (in relation to the original resolution).
147
+
For comparison of normal disparity vs. subpixel disparity images, click `here <https://github.com/luxonis/depthai/issues/184>`__.
100
148
101
-
Subpixel Disparity
102
-
******************
149
+
.. tab:: Mesh file / Homography matrix
103
150
104
-
Subpixel improves the precision and is especially useful for long range measurements. It also helps for better estimating surface normals
151
+
Mesh files are generated using the camera intrinsics, distortion coeffs, and rectification rotations.
152
+
These files helps in overcoming the distortions in the camera increasing the accuracy and also help in when `wide FOV <https://docs.luxonis.com/projects/hardware/en/latest/pages/arducam.html#arducam-compatible-cameras>`__ lens are used.
105
153
106
-
Besides the integer disparity output, the Stereo engine is programmed to dump to memory the cost volume, that is 96 levels (disparities) per pixel,
107
-
then software interpolation is done on Shave, resulting a final disparity with 5 fractional bits, resulting in significantly more granular depth
108
-
steps (32 additional steps between the integer-pixel depth steps), and also theoretically, longer-distance depth viewing - as the maximum depth
109
-
is no longer limited by a feature being a full integer pixel-step apart, but rather 1/32 of a pixel.
154
+
.. note::
155
+
Currently mesh files are generated only for stereo cameras on the host during calibration. The generated mesh files are stored in `depthai/resources <https://github.com/luxonis/depthai/tree/main/resources>`__ which users can load to the device. This process will be moved to on device in the upcoming releases.
- **Confidence threshold**: Stereo depth algorithm searches for the matching feature from right camera point to the left image (along the 96 dispairty levels). During this process it computes the cost for each disparity level and choses the minimal cost between two disparities and uses it to compute the confidence at each pixel. Stereo node will output disparity/depth pixels only where depth confidence is below the **confidence threshold** (lower the confidence value means better depth accuracy). Note: This threshold only applies to Normal stereo mode as of now.
172
+
- **LR check threshold**: Disparity is considered for the output when the difference between LR and RL disparities is smaller than the LR check threshold.
0 commit comments