You probably do not come from: United Arab Emirates. If necessary, change to: United States

O2D in detail

Contour detection vs. blob analysis

Contour verification

Contour detection is an important tool for 2D image processing. The edges as well as the transitions from foreground to background are detected and a contour is calculated from the information. The special feature of contour detection is that it also works reliably with interference caused by extraneous light, as the extraneous light usually hits the entire object. The relative difference between foreground and background shifts, but the contour is still detected with equal certainty. Object inspection is then performed by matching a reference contour with the current object.

Contour detection by means of:

  • Extraction of the object to be highlighted from the background by adjusting the lighting situation
  • Optimisation of the contour by deleting unnecessary areas
  • The algorithm detects possible contours in the live image that are distinguished as good or bad parts on the basis of a threshold value (score)

Where does contour detection apply:

The method is mainly used in pattern and shape detection as well as object recognition, as typically applied in punching, milling, turning or assembly. Contour detection is used for quality assurance in these areas.

Blob analysis

Blob analysis is an important image processing method in which image features are selected and analysed over a group of similar neighbouring pixels.

The BLOB (invented word Binary Large Object) in this context stands for Binary-Logic Data Object, which loosely translates as a set of pixels with the same logical state. The selection of the neighbouring pixels is generally done by thresholding the grey-scale value. Conclusions can then be drawn about various characteristics from the analysis. A well-known function is, e.g., the pixel counter.

Blob analysis by means of:

  • Extraction of the area of interest from the background by thresholding over the grey value
  • Optimisation of the search criteria via various attributes
  • Calculation of the characteristics searched for, such as number of pixels (pixel counter), area centre of gravity, orientation, shape (e.g. roundness, rectangularity) and diameter

Where does the blob analysis apply?

There are many different applications. For example, blob analysis can be used for monitoring completeness, detection of presence or thread detection as well as for counting and sorting objects.

Position tracking

The position tracking is done by means of an anchor contour that is found once in the image. Using this contour, search zones can track other models (for example, the search zone of a blob analysis) in position as well as orientation.
Graphical representation of a position tracking based on the example:

Detection of solder balls on a clip

  1. On the tips of a clip, it has to be checked whether all three solder balls are present (shown in green).
    Since the contour of a solder ball varies but the area of a solder ball remains constant, a blob analysis is used.The search zones shown in orange are defined for the presence monitoring over the area to be checked.
  2. In order to track these search zones depending on the position and orientation of the clip, a reference contour - the so-called anchor contour - is defined (shown in pink).The contour of the left rounding of the clip is then "anchored" with the search zones of the blob analysis.
  3. If the clip now rotates by 20 degrees, for example, the anchor contour is also found in the rotated state. The orange search zones of the blob analysis are then automatically tracked to the correct position and orientation.

CMOS processor

The O2D5 family from ifm uses a CMOS image processor with 1.2 MP (1280 x 960 pixels).

  • Each pixel contains a photon that collects and amplifies light from the camera lens.
  • Microlenses on each pixel maximise photon contact.
  • The photon accumulates an electrical charge proportional to the amount of light it receives.
  • The electrical charge is converted into an analogue voltage signal.
  • The analogue signal is transmitted to an A/D converter.
  • The image processor evaluates each digital signal and puts it together to an image.

CMOS image processors are easier, faster and cheaper to manufacture, making them the most widely used on the market.

LED lighting

In order to maximise contrast for each pixel it is important to choose the right illumination. The O2D family is supplied with integrated high-intensity LED light sources in RGB-W (red, green, blue, white) and infrared.

Please note that the image sensor is not a colour sensor!

However, choosing a light source with a different colour can have a dramatic effect on the contrast of the image. The picture below shows crayons in daylight and, for comparison, illuminated by the different LEDs of the O2D5 sensor.

efector dualis Lichtfarben

Comparison of the different light sources

Light type Please note:
  • The human eye recognises all colours equally in normal daylight.
  • Industrial inspection applications require features and contrasts of an object to be highlighted for evaluation by an image sensor.
Red light
  • Red portions appear brighter; colours like green and blue appear more contrasting and darker because they are absorbed.
  • Ideal for assessing printed objects (good contrast).
Green light
  • Green portions appear brighter; colours like red produce higher contrasts.
  • Ideal for the evaluation of metal objects.
Blue light
  • Blue portions appear brighter; colours such as red, yellow and green appear more contrasting.
  • Ideal for the evaluation of metal objects.
White light
  • White light contains all colours.
  • Ideal for distinguishing colours by their contrast (not absolute colour).
Infrared light
  • Daylight blocking filters allow the use of vision sensors independently of ambient light and compensate even for strongly changing light conditions or direct sunlight. The wavelength range is also crucial. For example, measurements in the infrared range are not exposed to the fluctuations of visible light.

Effect of the polarising filter

Due to reflections, it can be difficult to get sharp contours or areas on shiny objects. The O2D5 sensors with RGB-W light sources contain a polarisation filter that can be switched on or off to minimise the effect of reflections.

  1. without polarisation filter
  2. with polarisation filter