Version: 16.3.1.167ea6ecff

Peak Detector Tool
Note
The term peak used in this documentation refers to the maximum intensity of the laser stripe.

The software peak finder is a C++ library to extract the peak position of the laser stripe (see Acquisition Overview for a general description of laser detection). This library combined with 2D cameras can replaced the used of 3D cameras when the speed of the acquisition is not too important. The software peak finder is basically a tool to extract a profile for a given frame.

The following figure shows how the maximum light intensity position on the stripe can be extracted. The representation on the left shows the pixel values marked using bold dots, while the "continuous" intensity between discrete pixel values is shown in a dashed line. As can be seen, pixel values are nothing but samples of the real light intensity as it projects onto the image sensor. The higher is the sensor's resolution, the higher is the number of pixel values describing the intensity profile.

In this respect, it would be ideal to have a "continuous" image sensor, instead of a discrete pixel array, which is impractical for many reasons. Instead, we can use interpolation techniques in order to get a higher "interpolated" resolution along the direction perpendicular to the light stripe. In our case, we used a similar technique to 2nd degree interpolation between consecutive pixel values, obtaining interpolated values, as shown on the right side in the following figure.

peakDetector.png
Subpixel resolution detection of the maximum value position

The AQSENSE Peak Detector can determine the maximum intensity value position with an upper theoretical limit of 1/64th of a pixel. However, the detection accuracy is dependent on the image temporal noise, as well as laser spatial noise (speckle), surface reflection properties and reflectivity (the percentage of laser light reaching the sensor after projection onto an object).

SAL3D contains two different peak detectors, the horizontal peak finder and the vertical peak finder (obsolete). As the vertical peak finder is only on the SAL3D package for backward compatibility, the rest of the documentation is based only on the horizontal peak finder.

This peak detector delivers a single peak per column of the Frame. When multiple detections in a single column are given, the most significant among the detected is selected. The selection criteria involves the pixel intensity of each peak and its neighborhood, by selecting the stripe section with higher width x intensity product.

twoLaserOnARow.png
The AQSENSE peak detector detects the most significant peak in a column. The blue box shows the Frame region with two laser stripe sections, with the detection overlaid in red

A threshold on the pixel intensity value can be set to avoid the influence of the background pixels (those not belonging to the laser stripe). However, the higher robustness of this peak detector makes the detection almost independent of the careful selection of a threshold value.

peakThreshold.png
In this image a threshold of 40 will remove all points not belonging to the laser

Advantages of the Peak Detector tool

Acquisition Recommentations to achieve the best performance.

Comparing the peak with standard COG

Compared with the standard COG, the noise on the reconstruction using SAL3D peak detector is considerably smaller agains COG whitout loosing resolution on the detection. The following pictures display the reconstruction of an aluminium part.

full_reconstruction.png
Comparison SAL3D peak vs COG. a) Aluminum part; b) SAL3D reconstruction; c) COG recontruction.
zoom_reconstruction.png
Zoom reconstruction a) SAL3D; b) COG.