Back to Jetson Inference

Jetson Inference: Warping

docs/html/group__warping.html

latest9.9 KB
Original Source

| | Jetson Inference

DNN Vision Library |

Functions

Warping Utilities Library (jetson-utils) » CUDA

Various image warps and matrix transforms. More...

|

Functions

| | cudaError_t | cudaWarpAffine (uchar4 *input, uchar4 *output, uint32_t width, uint32_t height, const float transform[2][3], bool transform_inverted=false) | | | Apply 2x3 affine warp to an 8-bit fixed-point RGBA image. More...
| | | | cudaError_t | cudaWarpAffine (float4 *input, float4 *output, uint32_t width, uint32_t height, const float transform[2][3], bool transform_inverted=false) | | | Apply 2x3 affine warp to an 32-bit floating-point RGBA image. More...
| | | | cudaError_t | cudaWarpPerspective (void *input, uint32_t inputWidth, uint32_t inputHeight, imageFormat inputFormat, void *output, uint32_t outputWidth, uint32_t outputHeight, imageFormat outputFormat, const float transform[3][3], bool transform_inverted=false) | | | Apply a 3x3 perspective warp to an image. More...
| | | | template<typename T > | | cudaError_t | cudaWarpPerspective (T *input, uint32_t inputWidth, uint32_t inputHeight, T *output, uint32_t outputWidth, uint32_t outputHeight, const float transform[3][3], bool transform_inverted=false) | | | Apply a 3x3 perspective warp to an image. More...
| | | | cudaError_t | cudaWarpPerspective (uchar4 *input, uchar4 *output, uint32_t width, uint32_t height, const float transform[3][3], bool transform_inverted=false) | | | Apply 3x3 perspective warp to an 8-bit fixed-point RGBA image. More...
| | | | cudaError_t | cudaWarpPerspective (float4 *input, float4 *output, uint32_t width, uint32_t height, const float transform[3][3], bool transform_inverted=false) | | | Apply 3x3 perspective warp to an 32-bit floating-point RGBA image. More...
| | | | cudaError_t | cudaWarpIntrinsic (uchar4 *input, uchar4 *output, uint32_t width, uint32_t height, const float2 &focalLength, const float2 &principalPoint, const float4 &distortion) | | | Apply in-place instrinsic lens distortion correction to an 8-bit fixed-point RGBA image. More...
| | | | cudaError_t | cudaWarpIntrinsic (float4 *input, float4 *output, uint32_t width, uint32_t height, const float2 &focalLength, const float2 &principalPoint, const float4 &distortion) | | | Apply in-place instrinsic lens distortion correction to 32-bit floating-point RGBA image. More...
| | | | cudaError_t | cudaWarpFisheye (uchar4 *input, uchar4 *output, uint32_t width, uint32_t height, float focus) | | | Apply fisheye lens dewarping to an 8-bit fixed-point RGBA image. More...
| | | | cudaError_t | cudaWarpFisheye (float4 *input, float4 *output, uint32_t width, uint32_t height, float focus) | | | Apply fisheye lens dewarping to a 32-bit floating-point RGBA image. More...
| | |

Detailed Description

Various image warps and matrix transforms.

Function Documentation

cudaWarpAffine() [1/2]

| cudaError_t cudaWarpAffine | ( | float4 * | input, | | | | float4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | const float | transform[2][3], | | | | bool | transform_inverted = false | | | ) | | |

Apply 2x3 affine warp to an 32-bit floating-point RGBA image.

The 2x3 matrix transform is in row-major order (transform[row][column]) If the transform has already been inverted, set transform_inverted to true.

cudaWarpAffine() [2/2]

| cudaError_t cudaWarpAffine | ( | uchar4 * | input, | | | | uchar4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | const float | transform[2][3], | | | | bool | transform_inverted = false | | | ) | | |

Apply 2x3 affine warp to an 8-bit fixed-point RGBA image.

The 2x3 matrix transform is in row-major order (transform[row][column]) If the transform has already been inverted, set transform_inverted to true.

cudaWarpFisheye() [1/2]

| cudaError_t cudaWarpFisheye | ( | float4 * | input, | | | | float4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | float | focus | | | ) | | |

Apply fisheye lens dewarping to a 32-bit floating-point RGBA image.

Parameters

| [in] | focus | focus of the lens (in mm). |

cudaWarpFisheye() [2/2]

| cudaError_t cudaWarpFisheye | ( | uchar4 * | input, | | | | uchar4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | float | focus | | | ) | | |

Apply fisheye lens dewarping to an 8-bit fixed-point RGBA image.

Parameters

| [in] | focus | focus of the lens (in mm). |

cudaWarpIntrinsic() [1/2]

| cudaError_t cudaWarpIntrinsic | ( | float4 * | input, | | | | float4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | const float2 & | focalLength, | | | | const float2 & | principalPoint, | | | | const float4 & | distortion | | | ) | | |

Apply in-place instrinsic lens distortion correction to 32-bit floating-point RGBA image.

Pinhole camera model with radial (barrel) distortion and tangential distortion.

cudaWarpIntrinsic() [2/2]

| cudaError_t cudaWarpIntrinsic | ( | uchar4 * | input, | | | | uchar4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | const float2 & | focalLength, | | | | const float2 & | principalPoint, | | | | const float4 & | distortion | | | ) | | |

Apply in-place instrinsic lens distortion correction to an 8-bit fixed-point RGBA image.

Pinhole camera model with radial (barrel) distortion and tangential distortion.

cudaWarpPerspective() [1/4]

| cudaError_t cudaWarpPerspective | ( | float4 * | input, | | | | float4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | const float | transform[3][3], | | | | bool | transform_inverted = false | | | ) | | |

Apply 3x3 perspective warp to an 32-bit floating-point RGBA image.

The 3x3 matrix transform is in row-major order (transform[row][column]) If the transform has already been inverted, set transform_inverted to true.

**Deprecated:**this function has been deprecated in lieu of the overload that accepts different dimensions for the input and output images.

cudaWarpPerspective() [2/4]

template<typename T >

| cudaError_t cudaWarpPerspective | ( | T * | input, | | | | uint32_t | inputWidth, | | | | uint32_t | inputHeight, | | | | T * | output, | | | | uint32_t | outputWidth, | | | | uint32_t | outputHeight, | | | | const float | transform[3][3], | | | | bool | transform_inverted = false | | | ) | | |

Apply a 3x3 perspective warp to an image.

The 3x3 matrix transform is in row-major order (transform[row][column]) If the transform has already been inverted, set transform_inverted to true.

cudaWarpPerspective() [3/4]

| cudaError_t cudaWarpPerspective | ( | uchar4 * | input, | | | | uchar4 * | output, | | | | uint32_t | width, | | | | uint32_t | height, | | | | const float | transform[3][3], | | | | bool | transform_inverted = false | | | ) | | |

Apply 3x3 perspective warp to an 8-bit fixed-point RGBA image.

The 3x3 matrix transform is in row-major order (transform[row][column]) If the transform has already been inverted, set transform_inverted to true.

**Deprecated:**this function has been deprecated in lieu of the overload that accepts different dimensions for the input and output images.

cudaWarpPerspective() [4/4]

| cudaError_t cudaWarpPerspective | ( | void * | input, | | | | uint32_t | inputWidth, | | | | uint32_t | inputHeight, | | | | imageFormat | inputFormat, | | | | void * | output, | | | | uint32_t | outputWidth, | | | | uint32_t | outputHeight, | | | | imageFormat | outputFormat, | | | | const float | transform[3][3], | | | | bool | transform_inverted = false | | | ) | | |

Apply a 3x3 perspective warp to an image.

The 3x3 matrix transform is in row-major order (transform[row][column]) If the transform has already been inverted, set transform_inverted to true.

  • Generated on Tue Mar 28 2023 14:27:58 for Jetson Inference by 1.8.17