Warning

You are reading an old version of this documentation. If you want up-to-date information, please have a look at 1.1 .

4.2 Features Selection

Fig. 87 Pick Point Teaching - Features Selection

In this step, you must draw a mask on the image to highlight the relevant features of the part that should be used to position the pick point.

Use the in the toolbox (top right) to mark the visual features you want the system to track. These features will appear as cyan lines and will be used by EYE+ XTD to calculate the pick point position during production.

Tip

You can adjust the thickness of the tool with the slider. Use the eraser or CLEAR ALL if you need to reset your selection.

Not all features are relevant when defining the reference model for pick point positioning. The following guidelines will help you select only the most reliable and useful features:

• Same height as the pick point: Only include features that lie at the same height as the pick point. Features at different levels must not be selected as they can introduce positioning errors due to perspective projection. These features may shift relative to the pick point depending on the part’s position within the picking area.

• Reproducible across all candidates: The selected features must be consistent and repeatable on every instance of the part. Avoid including elements that may vary from one candidate to another, as this can reduce the reliability of the pick point detection.

• Sufficient features for unique identification: Ensure you provide enough distinct features so that the system can unambiguously identify the correct pick point. For example, if your pick point is in the center of a circular feature, but you omit the circle from the mask, the system might find multiple valid pick point positions.

  

  Fig. 88 Not enough features to position the pick points: 2 possibilities

Enter the part height

The part height you must enter in this step is the height where the pick point is placed.

EYE+ XTD corrects the perspective projection effect induced by the variation of the position of the parts on the picking area by adding an offset to the X and Y coordinates according to its position on the picking area. A part located on the edge of the picking area will have a greater offset than a part located in the middle.

To calculate these offsets, the system must identify the height of the part where you specified the pick point.

Fig. 89 Part height depending on the pick point position

Advanced options

The advanced options provide additional settings that can be adjusted in case the default settings do not give good enough results, for example when some important features are not detected.

Contrast

By default, EYE+ XTD automatically determines a contrast value that is used to extract features. The contrast value of a pixel is the gray value difference from its neighbors. If the pixel contrast exceeds the value specified, the pixel is selected to determine the model features.

Tweaking the contrast parameter allows creating a model that has more features than the automatically-computed one. For certain applications, it can help to precisely set the pick point.

The contrast setting can be manually set by toggling “Set contrast manually” on, as shown below:

The effect of the contrast setting is shown below:

Fig. 90 Pick point model where contrast is automatically computed by EYE+ XTD

Fig. 91 Pick point model where contrast is manually set to a lower value (e.g. 19)

Tip

The more features detected the more time it takes to run the pick angle step, remove as many features as possible using the mask functionality.

Scaling

EYE+ XTD Studio allows for uniform scaling of the model size from 0.5x to 1.5x in the pick point positioning step. This scaling feature is beneficial when the size of your parts varies uniformly, enabling the system to detect parts with different sizes and keep an accurate pick point.

Adjusting the scaling values allows for better model matching on the parts when your parts sizes varies a bit, making the pick point results more accurate. The scaling can be set differently during the pick point teaching and pick angle teaching steps to accommodate specific needs.

The scaling setting can be manually set by toggling “Set model scaling manually” on, as shown below:

When this option is enabled, the scaling factor will be displayed in the results table, indicating the scaling applied to parts that have been accepted.

The picture below represents a situation where the parts are rejected because the size of the model is not exactly the same. As a result, the algorithm does not recognize the features as identical and rejects the part.

Fig. 92 Results without manual scaling activated, showing the rejected parts that could have been accepted.

The picture below represents the same situation with the scaling feature activated. With the algorithm allowed to scale the features up and down, the parts are recognized correctly and accepted despite size variations.

Fig. 93 Results with manual scaling activated, showcasing the increased number of accepted parts.

Note

While increasing the scaling range can improve recognition and accuracy, it may also negatively impact computation time. To optimize performance, try removing as many features as possible using the mask functionality.

Note

The algorithm selects the closest match, and sometimes the accepted value may fall outside the defined range. This is due to the nature of the vision system and how the scaling operates to optimize feature recognition.

Warning

This feature is not intended for metrology purposes.