sb_t_surface_res Struct Reference

Defines the result of the Surface analysis. More...

#include <sb.h>

Collaboration diagram for sb_t_surface_res:

Data Fields
sb_t_point	ul
	Coordinates of the upper left corner.
sb_t_metrics	pixel_metrics
	Metrics calculated on predicted pixels. More...
sb_t_image *	model
	Model plane image. More...
sb_t_image *	truth
	Truth plane image. More...
sb_t_image *	weight
	Image of the voting plan. More...
sb_t_blobs *	blobs
	Defects blobs extracted. More...
sb_t_surface_image_truth	image_truth
	Truth of the image. More...

Detailed Description

Defines the result of the Surface analysis.

See sb_t_res for more information.

See also

sb_t_res

sb_project_get_res

How SB works

Definition at line 12352 of file sb.h.

Field Documentation

blobs

sb_t_blobs* sb_t_surface_res::blobs

Defects blobs extracted.

This information is filled only if the function sb_project_detection is called with the parameter roi_defects or if the parameter sb_t_par::surface_blob_analysis has been set to 1. sb_project_detection.
Each blob has the gray level code of the defect model.

Definition at line 12414 of file sb.h.

image_truth

sb_t_surface_image_truth sb_t_surface_res::image_truth

Truth of the image.

This is the global analysis result of the image

Definition at line 12420 of file sb.h.

model

sb_t_image* sb_t_surface_res::model

Model plane image.

Each pixel contain the gray level code of the model that has the greater weight. You can find the weight in the image sb_t_surface_res.weight.
The image format is SB_IMAGE_FORMAT_BW8.

Warning

The image is zero sized if the function sb_project_detection is called with a null or too small roi, or if a model is disabled.

Definition at line 12371 of file sb.h.

pixel_metrics

sb_t_metrics sb_t_surface_res::pixel_metrics

Metrics calculated on predicted pixels.

Model predicted for every pixel in the image is compared to the corresponding value in roi defects and the resulting truth values are used to compute metrics at pixel level. These statistics are useful to evaluate the overall quality of the detection without taking into account blobs .

Definition at line 12362 of file sb.h.

truth

sb_t_image* sb_t_surface_res::truth

Truth plane image.

Each pixel of the truth plane can be one of the following values:

• SB_TRUTH_TRUE_POSITIVE
• SB_TRUTH_TRUE_NEGATIVE
• SB_TRUTH_FALSE_POSITIVE
• SB_TRUTH_FALSE_NEGATIVE
• SB_TRUTH_OPTIONAL_POSITIVE
• SB_TRUTH_OPTIONAL_NEGATIVE The image format is SB_IMAGE_FORMAT_BW8.

  Warning

  The image is zero sized if the function sb_project_detection is called with a null or too small roi, or if a model is disabled.

Definition at line 12385 of file sb.h.

weight

sb_t_image* sb_t_surface_res::weight

Image of the voting plan.

The value of each pixel of the voting plane is the result of the prediction of the classificator. It is the weight or confidence of the defect. The range of value is from -1 to 1 quantized in 254 bins. So the values range is from 1 to 255, while 0 means undefined weight. The gray level 1 represents the value -1, the gray level 128 is 0, and the gray level 255 represents the +1.
There are two intervals:

• 1-127: negative values, pixel of a defect region
• 128-255: positive values, pixel of a good region

The image sb_t_surface_res::model contains the model of each pixel.
The image format is SB_IMAGE_FORMAT_BW8.
The SB GUI shows the voting plane using a jet map from blue (confidence -1) to red (confidence 1) as shown in the image below.

Voting plane

Warning

The image is zero sized if a invalid defects ROI (for example too small) is passed to the sb_project_detection routine or if a model is disabled.

Definition at line 12404 of file sb.h.

---

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

• sb.h