![]() |
VisionWorks Toolkit ReferenceDecember 18, 2015 | 1.2 Release |
The Semi-Global Matching Primitive is used for disparity map reconstruction.
The primitive uses the following identifiers:
vxGetKernelByEnum
function;"com.nvidia.nvx.semi_global_matching"
for vxGetKernelByName
function.Enumerations | |
enum | nvx_scanline_e { NVX_SCANLINE_LEFT_RIGHT = 1 << 0, NVX_SCANLINE_TOP_LEFT_BOTTOM_RIGHT = 1 << 1, NVX_SCANLINE_TOP_BOTTOM = 1 << 2, NVX_SCANLINE_TOP_RIGHT_BOTTOM_LEFT = 1 << 3, NVX_SCANLINE_RIGHT_LEFT = 1 << 4, NVX_SCANLINE_BOTTOM_RIGHT_TOP_LEFT = 1 << 5, NVX_SCANLINE_BOTTOM_TOP = 1 << 6, NVX_SCANLINE_BOTTOM_LEFT_TOP_RIGHT = 1 << 7, NVX_SCANLINE_CROSS, NVX_SCANLINE_ALL = 0xFF } |
Defines scan line's directions used during cost aggregation step. More... | |
Functions | |
vx_node | nvxSemiGlobalMatchingNode (vx_graph graph, vx_image left, vx_image right, vx_image disparity, vx_int32 minD, vx_int32 maxD, vx_int32 P1, vx_int32 P2, vx_int32 sad, vx_int32 clip, vx_int32 max_diff, vx_int32 uniqueness, vx_enum scanlines_mask) |
[Graph] evaluate disparity given 2 stereo images using the SGM algorithm. More... | |
vx_status | nvxuSemiGlobalMatching (vx_context context, vx_image left, vx_image right, vx_image disparity, vx_int32 minD, vx_int32 maxD, vx_int32 P1, vx_int32 P2, vx_int32 sad, vx_int32 clip, vx_int32 max_diff, vx_int32 uniqueness, vx_enum scanlines_mask) |
[Immediate] evaluate disparity given 2 stereo images using the SGM algorithm. More... | |
enum nvx_scanline_e |
Defines scan line's directions used during cost aggregation step.
vx_node nvxSemiGlobalMatchingNode | ( | vx_graph | graph, |
vx_image | left, | ||
vx_image | right, | ||
vx_image | disparity, | ||
vx_int32 | minD, | ||
vx_int32 | maxD, | ||
vx_int32 | P1, | ||
vx_int32 | P2, | ||
vx_int32 | sad, | ||
vx_int32 | clip, | ||
vx_int32 | max_diff, | ||
vx_int32 | uniqueness, | ||
vx_enum | scanlines_mask | ||
) |
[Graph] evaluate disparity given 2 stereo images using the SGM algorithm.
[in] | graph | Specifies the graph. |
[in] | left | Left stereo image |
[in] | right | Right stereo image |
[out] | disparity | Resulting "disparity" map. It contains values from minD-1 to maxD (exclusive) represented in Q11.4 fixed point format. |
[in] | minD | The minimum disparity value. |
[in] | maxD | The maximum disparity value. |
[in] | P1 | Smoothness penalty. |
[in] | P2 | Discontiguous penalty. |
[in] | sad | Specifies the average window size for sum of absolute differences (set to 1 for original SGM). |
[in] | clip | Clip value for cost (used to preserve overflow in cost function) |
[in] | max_diff | Maximum allowed difference (in integer pixel units) in the left-right disparity check. |
[in] | uniqueness | Margin in percentage by which the best (minimum) computed cost function value should “win” the second best value to consider the found match correct. |
[in] | scanlines_mask | Scan line directions. Any bit-OR of nvx_scanline_e values is suitable. |
vx_status nvxuSemiGlobalMatching | ( | vx_context | context, |
vx_image | left, | ||
vx_image | right, | ||
vx_image | disparity, | ||
vx_int32 | minD, | ||
vx_int32 | maxD, | ||
vx_int32 | P1, | ||
vx_int32 | P2, | ||
vx_int32 | sad, | ||
vx_int32 | clip, | ||
vx_int32 | max_diff, | ||
vx_int32 | uniqueness, | ||
vx_enum | scanlines_mask | ||
) |
[Immediate] evaluate disparity given 2 stereo images using the SGM algorithm.
[in] | context | Specifies the context. |
[in] | left | Left stereo image |
[in] | right | Right stereo image |
[out] | disparity | Resulting "disparity" map. It contains values from minD-1 to maxD (exclusive) represented in Q11.4 fixed point format. |
[in] | minD | The minimum disparity value. |
[in] | maxD | The maximum disparity value. |
[in] | P1 | Smoothness penalty. |
[in] | P2 | Discontiguous penalty. |
[in] | sad | Specifies the average window size for sum of absolute differences (set to 1 for original SGM). |
[in] | clip | Clip value for cost (used to preserve overflow in cost function) |
[in] | max_diff | Maximum allowed difference (in integer pixel units) in the left-right disparity check. |
[in] | uniqueness | Margin in percentage by which the best (minimum) computed cost function value should “win” the second best value to consider the found match correct. |
[in] | scanlines_mask | Scan line directions. Any bit-OR of nvx_scanline_e values is suitable. |