F_VN_Threshold

Syntax

Definition:

FUNCTION F_VN_Threshold : HRESULT
VAR_INPUT
    ipSrcImage     : ITcVnImage;
    ipDestImage    : Reference To  ITcVnImage;
    fThreshold     : LREAL;
    fMaxValue      : LREAL;
    eThresholdType : ETcVnThresholdType;
    hrPrev         : HRESULT;
END_VAR

 Inputs

 Return value

HRESULT

Further information

The function F_VN_Threshold segments the input image with regard to a threshold. Depending on this threshold and the algorithm used, each pixel of the binary result image is assigned the value 0 and a specified target value.

Algorithm

A threshold transforms a monochrome image into a binary image and thus creates segments. Good segmentation is a prerequisite for many subsequent methods such as contour search. TwinCAT Vision offers different types of threshold values that are suitable for different applications.

Fixed threshold value

A fixed threshold simply creates a binary image by setting pixels with a value greater than fThreshold to white (255) and others to black (0). fThreshold is a parameter of the corresponding functions. To find a good value, you will most probably approach it via different attempts.

Dynamic threshold value

A dynamic threshold according to Otsu works perfectly for bimodal images. The histogram of a bimodal image has two main peaks and a good threshold value is most likely exactly in the middle. If you have a noisy image, you can use a Gaussian filter to shift it towards bimodal.

Adaptive threshold value

An adaptive threshold value uses variable threshold values for the whole image. See F_VN_AdaptiveThresholdExp .

Parameter

Input image

The input image ipSrcImage may have one or three channels.

Result image

The result image ipDestImage returns the segmented binary image.

Threshold value

The threshold fThreshold defines the segmentation condition. It must be adjusted depending on the properties of the input image.

Target value

The target value fMaxValue defines which value should be given to the pixels that fulfil the threshold criterion. If the result image is to consist only of black and white pixels, then the maximum value of the element type of the input image must be set here. This corresponds to (2^d)-1, where d represents the bit depth of the element type; e.g. (2^8)-1=255 with 8-bit images.

Threshold type

The threshold type eThresholdType defines the algorithm according to which the threshold fThreshold is applied to the input image. All values of the enum ETcVnThresholdType are supported.

Application

The binary segmentation of an 8-bit grayscale image in the center of the value range looks like this, for example:

hr := F_VN_Threshold(
    ipSrcImage      :=  ipImageIn,
    ipDestImage     :=  ipImageRes,
    fThreshold      :=  128, // half value range: (2^8)/2=128
    fMaxValue       :=  255, // max value of 8-bit image: (2^8)-1=255
    eThresholdType  :=  TCVN_TT_BINARY,
    hrPrev          :=  hr
);

Required License

TC3 Vision Base

System Requirements