StereoDepthConfig

Align the disparity/depth to the perspective of a rectified output, or center it  Members:    RECTIFIED_RIGHT :     RECTIFIED_LEFT :     CENTER : 

Measurement unit for depth data  Members:    METER :     CENTIMETER :     MILLIMETER :     INCH :     FOOT :     CUSTOM : 

Used only for debug purposes. centerAlignmentShiftFactor is set automatically in firmware, from camera extrinsics when depth alignment to camera is enabled. Center alignment is achieved by shifting the obtained disparity map by a scale factor. It's used to align to a different camera that is on the same horizontal baseline as the two stereo cameras. E.g. if we have a device with 10 cm stereo baseline, and we have another camera inbetween, 9cm from the LEFT camera and 1 cm from the RIGHT camera we can align the obtained disparity map using a scale factor of 0.9. Note that aligning disparity map to a different camera involves 2 steps: 1. Shifting obtained disparity map. 2. Warping the image to counter rotate and scaling to match the FOV. Center alignment factor 1 is equivalent to RECTIFIED_RIGHT Center alignment factor 0 is equivalent to RECTIFIED_LEFT

Custom depth unit multiplier, if custom depth unit is enabled, relative to 1 meter. A multiplier of 1000 effectively means depth unit in millimeter.

Set the disparity/depth alignment to the perspective of a rectified output, or center it

Measurement unit for depth data. Depth data is integer value, multiple of depth unit.

Shift input frame by a number of pixels to increase minimum depth. For example shifting by 48 will change effective disparity search range from (0,95] to [48,143]. An alternative approach to reducing the minZ. We normally only recommend doing this when it is known that there will be no objects farther away than MaxZ, such as having a depth camera mounted above a table pointing down at the table surface.

Disparity range increased from 95 to 190, combined from full resolution and downscaled images. Suitable for short range objects

Computes and combines disparities in both L-R and R-L directions, and combine them. For better occlusion handling

Computes disparity with sub-pixel interpolation (5 fractional bits), suitable for long range

Left-right check threshold for left-right, right-left disparity map combine, 0..128 Used only when left-right check mode is enabled. Defines the maximum difference between the confidence of pixels from left-right and right-left confidence maps

Invalidate X amount of pixels at the edge of disparity frame. For right and center alignment X pixels will be invalidated from the right edge, for left alignment from the left edge.

Number of fractional bits for subpixel mode  Valid values: 3,4,5  Defines the number of fractional disparities: 2^x  Median filter postprocessing is supported only for 3 fractional bits

Census transform kernel size possible values.  Members:    AUTO :     KERNEL_5x5 :     KERNEL_7x7 :     KERNEL_7x9 : 

If enabled, each pixel in the window is compared with the mean window value instead of the central pixel.

Census transform mask, default - auto, mask is set based on resolution and kernel size. Disabled for 400p input resolution. Enabled for 720p. 0XA82415 for 5x5 census transform kernel. 0XAA02A8154055 for 7x7 census transform kernel. 0X2AA00AA805540155 for 7x9 census transform kernel. Empirical values.

Census transform kernel size.

Census transform comparison threshold value.

Cost calculation linear equation parameters.

Horizontal P1 penalty cost parameter.

Horizontal P2 penalty cost parameter.

Vertical P1 penalty cost parameter.

Vertical P2 penalty cost parameter.

Disparity search range: 64 or 96 pixels are supported by the HW.  Members:    DISPARITY_64 :     DISPARITY_96 : 

The linear equation applied for computing the cost is: COMB_COST = α*AD + β*(CTC<<3). CLAMP(COMB_COST >> 5, threshold). Where AD is the Absolute Difference between 2 pixels values. CTC is the Census Transform Cost between 2 pixels, based on Hamming distance (xor). The α and β parameters are subject to fine tuning by the user.

Disparities with confidence value under this threshold are accepted. Higher confidence threshold means disparities with less confidence are accepted too.

Disparity search range, default 96 pixels.

Disparity companding using sparse matching. Matching pixel by pixel for N disparities. Matching every 2nd pixel for M disparitites. Matching every 4th pixel for T disparities. In case of 96 disparities: N=48, M=32, T=16. This way the search range is extended to 176 disparities, by sparse matching. Note: when enabling this flag only depth map will be affected, disparity map is not.

Used only for debug purposes, SW postprocessing handled only invalid value of 0 properly.

Cost calculation linear equation parameters.

Brightness filtering. If input frame pixel is too dark or too bright, disparity will be invalidated. The idea is that for too dark/too bright pixels we have low confidence, since that area was under/over exposed and details were lost.

Decimation filter. Reduces the depth scene complexity. The filter runs on kernel sizes [2x2] to [8x8] pixels.

1D edge-preserving spatial filter using high-order domain transform.

Speckle filtering. Removes speckle noise.

Temporal filtering with optional persistence.

Threshold filtering. Filters out distances outside of a given interval.

Sigma value for bilateral filter. 0 means disabled. A larger value of the parameter means that farther colors within the pixel neighborhood will be mixed together.

Brightness filtering. If input frame pixel is too dark or too bright, disparity will be invalidated. The idea is that for too dark/too bright pixels we have low confidence, since that area was under/over exposed and details were lost.

Decimation filter. Reduces disparity/depth map x/y complexity, reducing runtime complexity for other filters.

Set kernel size for disparity/depth median filtering, or disable

Edge-preserving filtering: This type of filter will smooth the depth noise while attempting to preserve edges.

Speckle filtering. Removes speckle noise.

Temporal filtering with optional persistence.

Threshold filtering. Filters out distances outside of a given interval.