_vx_imagepatch_addressing_t Struct Reference

Data Fields

_vx_imagepatch_addressing_t Struct Reference

The addressing image patch structure is used by the Host only to address pixels in an image patch. The fields of the structure are defined as: More...

#include <develop/amd_openvx/openvx/include/VX/vx_types.h>

Data Fields
vx_uint32	dim_x
	Width of patch in X dimension in pixels.
vx_uint32	dim_y
	Height of patch in Y dimension in pixels.
vx_int32	stride_x
	Stride in X dimension in bytes.
vx_int32	stride_y
	Stride in Y dimension in bytes.
vx_uint32	scale_x
	Scale of X dimension. For sub-sampled planes this is the scaling factor of the dimension of the plane in relation to the zero plane. Use VX_SCALE_UNITY in the numerator.
vx_uint32	scale_y
	Scale of Y dimension. For sub-sampled planes this is the scaling factor of the dimension of the plane in relation to the zero plane. Use VX_SCALE_UNITY in the numerator.
vx_uint32	step_x
	Step of X dimension in pixels.
vx_uint16	step_y
	Step of Y dimension in pixels.
vx_uint16	stride_x_bits
	Stride in X dimension in bits. Used when stride_x is not an integer number of bytes (e.g. for VX_DF_IMAGE_U1 images).

Detailed Description

The addressing image patch structure is used by the Host only to address pixels in an image patch. The fields of the structure are defined as:

• dim - The dimensions of the image in logical pixel units in the x & y direction.
• stride - The physical byte distance from a logical pixel to the next logically adjacent pixel in the positive x or y direction.
• scale - The relationship of scaling from the primary plane (typically the zero indexed plane) to this plane. An integer down-scaling factor of \( f \) shall be set to a value equal to \( scale = \frac{unity}{f} \) and an integer up-scaling factor of \( f \) shall be set to a value of \( scale = unity * f \). \( unity \) is defined as VX_SCALE_UNITY.
• step - The step is the number of logical pixel units to skip to arrive at the next physically unique pixel. For example, on a plane that is half-scaled in a dimension, the step in that dimension is 2 to indicate that every other pixel in that dimension is an alias. This is useful in situations where iteration over unique pixels is required, such as in serializing or de-serializing the image patch information.
• stride_x_bits - The physical bit distance from a logical pixel to the next logically adjacent pixel in the positive x-direction. This field is only used when the stride in the x-direction is not an integer number of bytes (e.g. for VX_DF_IMAGE_U1 images).

  See also

  vxMapImagePatch

  Note

  For VX_DF_IMAGE_U1 images it is defined that stride_x == 0 since it is less than one byte. The least significant bit (bit number 0) in the first byte in the image, is the left-most pixel in the upper left corner, i.e. origin. A VX_DF_IMAGE_U1 image always starts on a byte boundary and each row has a stride_y that is a multiple of whole bytes, which means padding bits of implementation-defined values may be present at the end of each row. Image patches can only be accessed at a multiple of eight pixels: the x-coordinate must be a multiple of eight. Individual pixel access is also different: the byte at the imagepatch-calculated pointer value is a collection of eight pixels. Each byte can then be masked with the bit-mask 1 << (x % 8) to get individual pixel values (shifted x times). See sub_image_access for an example of pixel addressing on a VX_DF_IMAGE_U16 image.

  If data type is contained in a whole number of bytes, then stride_x MUST be equal to the number of bytes represented by that data type (e.g. For VX_DF_IMAGE_U16, stride_x must be 2 only).

---

The documentation for this struct was generated from the following file:

• develop/amd_openvx/openvx/include/VX/vx_types.h