Skip to content

blaize::isp

Types

Name
enum class WhiteBalanceMode
enum class LensDistortionMode
enum class LensDistortionZoomType
enum class Interpolationtype
enum class BayerPacking

Functions

Name
vx_node BlackLevelCorrectionNode(vx_graph graph, vx_image input, vx_image output, vx_pixel_value_t black_level)
Adjusts the displayed black level to black level information of image signal.
vx_node BadPixelCorrectionNode(vx_graph graph, vx_image input, vx_image output)
Removes defective pixels like dead-pixels, hot-pixels, bright-pixel and black-pixels.
vx_node DemosaicNode(vx_graph graph, vx_image input, vx_image rgb_out)
Reproduces the complete color image from the partial color image data, received from a Single-Sensor Camera or an image.
vx_node GammaCorrectionNode(vx_graph graph, vx_image rgb_in, vx_image rgb_out, vx_float32 gamma)
Runs the gamma correction function on input image to correct its luminance output(x,y) = (input(x,y))^(gamma)
vx_node ColorCorrectionNode(vx_graph graph, vx_image rgb_in, vx_image rgb_out, vx_tensor correction_matrix)
this module improves the RGB data to provide a faithful color rendition
vx_node WhiteBalanceNode(vx_graph graph, vx_image rgb_in, vx_image rgb_out, vx_float32 red_gain, vx_float32 green_gain, vx_float32 blue_gain)
Adjusts the coloration of images captured under different illuminations.
vx_node CropImageNode(vx_graph graph, vx_image input, vx_image output, vx_coordinates2d_t top_left)
Crops given bayer image.
vx_node AutoWhiteBalanceNode(vx_graph graph, vx_image rgb_in, vx_image rgb_out, WhiteBalanceMode awb_mode)
AWB estimate accurately the colour of the overall scene illumination and to make the image look as if is taken under canonical light.
vx_node DenoiseNode(vx_graph graph, vx_image yuv_in, vx_image yuv_out, vx_float32 soft_thres, vx_float32 hard_thres)
Denoise Node, reduces noise in an image.
vx_node EdgeEnhancementNode(vx_graph graph, vx_image input, vx_image output)
Enhances the edge contrast of an image to improve its sharpness.
vx_node LensDistortionNode(vx_graph graph, vx_image input, vx_image output, LensDistortionMode type, LensDistortionZoomType zoom_type, vx_float32 strength)
This module fixes the deviation from rectilinear projection caused due to the lens used.
vx_node WDRToneMappingNode(vx_graph graph, vx_image yuv_in, vx_image yuv_out)
WDR Wide dynamic range tone mapping module. This module obtains a perceptual match between a real scene and a displayed image.
vx_node DehazingNode(vx_graph graph, vx_image yuv_in, vx_image yuv_out, vx_float32 trans_gain)
Removes the haze due to atmosphere, which degrades the quality of images. Note: This is also referred to as dehazing in Python documents and code.
vx_node ScaleImageNode(vx_graph graph, vx_image input, vx_image output, Interpolationtype interpolation)
Scales down the given input bayer or RGB image with given interpolation.

Types Documentation

enum WhiteBalanceMode

Enumerator Value Description
WHITE_WORLD 0
GREY_WORLD

enum LensDistortionMode

Enumerator Value Description
PINCUSHION_1 0
PINCUSHION_2
BARREL_1
BARREL_2

enum LensDistortionZoomType

Enumerator Value Description
FIT 0
CROP

enum Interpolationtype

Enumerator Value Description
VX_INTERPOLATION_NEAREST_NEIGHBOR 0
VX_INTERPOLATION_BILINEAR

enum BayerPacking

Enumerator Value Description
RGGB 0
RGBG
BGGR
GRGB
INVALID

Functions Documentation

function BlackLevelCorrectionNode

vx_node BlackLevelCorrectionNode(
    vx_graph graph,
    vx_image input,
    vx_image output,
    vx_pixel_value_t black_level
)

Adjusts the displayed black level to black level information of image signal.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] input bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16
  • output [OUT] output bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16
  • black_level [IN] the pixel value corresponds to the black, which is used to correct pixels.

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function BadPixelCorrectionNode

vx_node BadPixelCorrectionNode(
    vx_graph graph,
    vx_image input,
    vx_image output
)

Removes defective pixels like dead-pixels, hot-pixels, bright-pixel and black-pixels.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] input bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16
  • output [OUT] output bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function DemosaicNode

vx_node DemosaicNode(
    vx_graph graph,
    vx_image input,
    vx_image rgb_out
)

Reproduces the complete color image from the partial color image data, received from a Single-Sensor Camera or an image.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] input bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16
  • rgb_out [OUT] a vx_image of type VX_DF_IMAGE_RGB and VX_DF_IMAGE_U16

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function GammaCorrectionNode

vx_node GammaCorrectionNode(
    vx_graph graph,
    vx_image rgb_in,
    vx_image rgb_out,
    vx_float32 gamma
)

Runs the gamma correction function on input image to correct its luminance output(x,y) = (input(x,y))^(gamma)

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • rgb_in [IN] a vx_image of type VX_DF_IMAGE_RGB
  • rgb_out [OUT] a vx_image of type VX_DF_IMAGE_RGB
  • g_error [IN] gamma error value of type vx_float32.

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function ColorCorrectionNode

vx_node ColorCorrectionNode(
    vx_graph graph,
    vx_image rgb_in,
    vx_image rgb_out,
    vx_tensor correction_matrix
)

this module improves the RGB data to provide a faithful color rendition

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • rgb_in [IN] a vx_image of type VX_DF_IMAGE_RGB
  • rgb_out [OUT] a vx_image of type VX_DF_IMAGE_RGB
  • correction_matrix [IN] Expects a vx_tensor array of shape (3, 3) containing ccm data

Return:

  • vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.
  • vx_node on a successful exection the API returns a valid openvx node else it returns NULL.

function WhiteBalanceNode

vx_node WhiteBalanceNode(
    vx_graph graph,
    vx_image rgb_in,
    vx_image rgb_out,
    vx_float32 red_gain,
    vx_float32 green_gain,
    vx_float32 blue_gain
)

Adjusts the coloration of images captured under different illuminations.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • rgb_in [IN] a vx_image of type VX_DF_IMAGE_RGB
  • rgb_out [OUT] a vx_image of type VX_DF_IMAGE_RGB
  • awb_mode [IN] auto white balance of type isp_wb_mode_e. currently supported awb_mode WHITE_WORLD, GREY_WORLD
  • red_gain [IN] white balance red gain value of type vx_float32.
  • green_gain [IN] white balance green gain value of type vx_float32.
  • blue_gain [IN] white balance blue gain value of type vx_float32.

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function CropImageNode

vx_node CropImageNode(
    vx_graph graph,
    vx_image input,
    vx_image output,
    vx_coordinates2d_t top_left
)

Crops given bayer image.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] input bayer or RGB image. currently supported vx_image of type VX_DF_IMAGE_U8 VX_DF_IMAGE_U16, VX_DF_IMAGE_RGB, TC_DF_IMAGE_RGGB_U8, TC_DF_IMAGE_RGBG_U8, TC_DF_IMAGE_GRGB_U8, TC_DF_IMAGE_BGGR_U8, TC_DF_IMAGE_RGGB_U16, TC_DF_IMAGE_RGBG_U16, TC_DF_IMAGE_GRGB_U16, TC_DF_IMAGE_BGGR_U16
  • output [OUT] output bayer or RGB image. currently supported vx_image of type VX_DF_IMAGE_U8 VX_DF_IMAGE_U16, VX_DF_IMAGE_RGB, TC_DF_IMAGE_RGGB_U8, TC_DF_IMAGE_RGBG_U8, TC_DF_IMAGE_GRGB_U8, TC_DF_IMAGE_BGGR_U8, TC_DF_IMAGE_RGGB_U16, TC_DF_IMAGE_RGBG_U16, TC_DF_IMAGE_GRGB_U16, TC_DF_IMAGE_BGGR_U16. The output type must match the input type.
  • top_left [IN] top left coordinate for cropping, the width and height are determined from the shape of the output image.

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function AutoWhiteBalanceNode

vx_node AutoWhiteBalanceNode(
    vx_graph graph,
    vx_image rgb_in,
    vx_image rgb_out,
    WhiteBalanceMode awb_mode
)

AWB estimate accurately the colour of the overall scene illumination and to make the image look as if is taken under canonical light.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • rgb_in [IN] a vx_image of type VX_DF_IMAGE_RGB
  • rgb_out [OUT] a vx_image of type VX_DF_IMAGE_RGB
  • awb_mode [IN] auto white balance of type isp_wb_mode_e. currently supported awb_mode WHITE_WORLD, GREY_WORLD

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function DenoiseNode

vx_node DenoiseNode(
    vx_graph graph,
    vx_image yuv_in,
    vx_image yuv_out,
    vx_float32 soft_thres,
    vx_float32 hard_thres
)

Denoise Node, reduces noise in an image.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • yuv_in [IN] a vx_image of type VX_DF_IMAGE_YUV4
  • yuv_out [OUT] a vx_image of type VX_DF_IMAGE_YUV4
  • soft_thres [IN] soft_thres value of type vx_float32.
  • hard_thres [IN] hard_thres value of type vx_float32.

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function EdgeEnhancementNode

vx_node EdgeEnhancementNode(
    vx_graph graph,
    vx_image input,
    vx_image output
)

Enhances the edge contrast of an image to improve its sharpness.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] a vx_image of type VX_DF_IMAGE_RGB
  • output [OUT] a vx_image of type VX_DF_IMAGE_RGB

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function LensDistortionNode

vx_node LensDistortionNode(
    vx_graph graph,
    vx_image input,
    vx_image output,
    LensDistortionMode type,
    LensDistortionZoomType zoom_type,
    vx_float32 strength
)

This module fixes the deviation from rectilinear projection caused due to the lens used.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] input bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16
  • output [OUT] output bayer image. currently supported vx_image of type VX_DF_IMAGE_U8 and VX_DF_IMAGE_U16
  • type [IN] type of correction to be done. LensDistortionMode currently supported types PINCUSHION_1, PINCUSHION_2, BARREL_1, BARREL_2
  • zoom_type [IN] type of zoom to be applied. LensDistortionZoomType supported types FIT, CROP
  • strength [IN] strength value for lens distortion of type vx_float32

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function WDRToneMappingNode

vx_node WDRToneMappingNode(
    vx_graph graph,
    vx_image yuv_in,
    vx_image yuv_out
)

WDR Wide dynamic range tone mapping module. This module obtains a perceptual match between a real scene and a displayed image.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • yuv_in [IN] a vx_image of type VX_DF_IMAGE_YUV4
  • yuv_out [OUT] a vx_image of type VX_DF_IMAGE_YUV4

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.

function DehazingNode

vx_node DehazingNode(
    vx_graph graph,
    vx_image yuv_in,
    vx_image yuv_out,
    vx_float32 trans_gain
)

Removes the haze due to atmosphere, which degrades the quality of images. Note: This is also referred to as dehazing in Python documents and code.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the Dehazing module's node have to be created
  • yuv_in [IN] a vx_image of type VX_DF_IMAGE_YUV4
  • yuv_out [OUT] a vx_image of type VX_DF_IMAGE_YUV4
  • trans_gain [IN] trans_gain value of type vx_float32.

Return: vx_node  on a successful execution the API returns a valid openvx node else it returns  *          a node in error state, use vxGetStatus to check the status of the returned node, and  *          vxGetLastErrorMessage.

function ScaleImageNode

vx_node ScaleImageNode(
    vx_graph graph,
    vx_image input,
    vx_image output,
    Interpolationtype interpolation
)

Scales down the given input bayer or RGB image with given interpolation.

Parameters:

  • graph [IN] OpenVX graph object (vx_graph) on which the node will be created
  • input [IN] input bayer or RGB image. Supports vx_image of type VX_DF_IMAGE_U8 VX_DF_IMAGE_U16, VX_DF_IMAGE_RGB, TC_DF_IMAGE_RGGB_U8, TC_DF_IMAGE_RGBG_U8, TC_DF_IMAGE_GRGB_U8, TC_DF_IMAGE_BGGR_U8, TC_DF_IMAGE_RGGB_U16, TC_DF_IMAGE_RGBG_U16, TC_DF_IMAGE_GRGB_U16, TC_DF_IMAGE_BGGR_U16
  • output [OUT] output bayer or RGB image. currently supported vx_image of type VX_DF_IMAGE_U8 VX_DF_IMAGE_U16, VX_DF_IMAGE_RGB, TC_DF_IMAGE_RGGB_U8, TC_DF_IMAGE_RGBG_U8, TC_DF_IMAGE_GRGB_U8, TC_DF_IMAGE_BGGR_U8, TC_DF_IMAGE_RGGB_U16, TC_DF_IMAGE_RGBG_U16, TC_DF_IMAGE_GRGB_U16, TC_DF_IMAGE_BGGR_U16. The output type must match the input type.
  • interpolation [IN] interpolation type for scaling of type vx_interpolation_type_e. currently supported interpolation types VX_INTERPOLATION_NEAREST_NEIGHBOR

Return: vx_node on a successful execution the API returns a valid openvx node else it returns a node in error state, use vxGetStatus to check the status of the returned node, and vxGetLastErrorMessage.


Updated on 2023-11-29 at 18:22:22 +0000