Dyn_threshold (OrigImage ThresholdImage: RegionDynThresh: Offset, LightDark:)

The contrast between the background, you can set the global threshold threshold, but in many cases, because of the unequal background,
Target body often characterized by local brighter or darker than the background, unable to determine the global threshold operation, need through the neighborhood to find a suitable threshold segmentation,
ThresholdImage is a reference image, compared with OrigImage find areas to determine the threshold, the commonly used smoothing filter operator (such as mean_image) for reference images,
Set offset range of neighborhood comparison, grey value change within the scope of the offset is acceptable,
Make g_ {o}=g_ {OrigImage},
G_ {t}=g_ {ThresholdImage}
Represent the gray levels of pixels of the original image and the reference map.
Our approach is to put the reference image grey value plus an Offset (minus),
Then go to and the original image pixel by pixel corresponding to compare,
See below for these formulas given in the Halcon:

The condition for? LightDark? Is:='light' g_o & gt;=g_t + offset, light extract relative reference figure brighter areas,


For LightDark='dark', the conditon is:? G_o & lt;=g_t - offset, dark extract relative reference figure darker areas of


For LightDark='equal', g_t - offset<=g_o & lt;=g_t + offset, the selection and reference about the same area,




Shape_trans (Region: RegionTrans: Type:)
Function:
The shape of the transformation zone

Parameter Type optional explanation is as follows:
Convex: convex hull
The ellipse, and have the same torque input area and area of the elliptical
Outer_circle: minimum circumscribed circle
Inner_circle: maximum inscribed circle
Rectangle1: parallel to the axis of the minimum circumscribed rectangle
Rectangle2: minimum circumscribed rectangle
Inner_rectangle1: parallel to the axis of maximum inscribed rectangle
Inner_rectangle2: input area of skeleton point with same input area centre of the smallest distance