CALIBRATE AND MEASURE EXTENSIONS TO FOIL
This software suite can be used to calibrate the camera field and measure objects in the camera field of view. The calibrate section lists the routines used to calibrate the camera image as downloaded from the framegrabber. The measure section allows the user to the measure objects in these images.
Dark Field Subtraction
Used to balance the "black" values from the camera
Imaging the inspection area without a part in it typically provides the data for this. This measurement is used to develop thresholds for later measurements.
Ideally a calibration standard is provided, such as an array of dots, which span the whole view area of the camera. This allows the calculation of scale and variations in scale across the field of view. Three corrections are typically performed,
- Perspective correction,
- Color Aberration correction, and
- Scale correction.
Raw Image Coordinates
(ix,iy) these are the coordinates of each pixel in the image. The coordinate system origin is the upper left hand corner of the image, x extends to the right, y extends down.
Corrected Image Coordinates
(cx,cy) these are the coordinates of each pixel in the corrected image, after Perspective, Color aberration, and scale correction. The origin is the center of the image, with x extending to the right, and y extending up.
Object coordinates are used to place measurement tools and report measurements. Object coordinates are reached by an affine transformation from corrected image coordinates.
An initial step in measuring an image is to determine the orientation of the object in the field of view. Locating some prominent features of the object and setting the origin and angle of the world coordinate system to match the orientation of the object to be measured do this.
One method of locating the object is to compute its minimum-area-bounding box. This algorithm finds the smallest box that can contain the image of the object. The software can then use the long side of the box to orient the system.
- Find the centroid of the image of the object.
- Measure the diameter of the object along radials from the centroid.
- Take the maximum and the minimum diameters. Pick the bounding box with less area determined by these two diameters.
Sometimes the bounding box is not enough, for example when the part is square. Locating a hole often can remove the ambiguity.
- Collect a set of points that have no part with their convex hull.
- Compute the center of the convex hull
Finding a line of the required length can also be used to orient the object.
- From a collection of points partition the points in to sets, which contain more than four points that are co-linear (with tolerance).
- Eliminate sets whose lines cross the interior of the part.
- Extend the sets to the end points.
Place the intersections of lines in the object. Select a corner pair as the base line of the object. The outline of the Algorithm:
- Locate all the lines in the part, and extend them to their endpoints.
- All the end points of the lines are at corners.
Image Alignment Tools
After the orientation of the part is determined, an affine transformation is defined which moves the part into the ‘standard’ position and orientation for measurement. This typically requires, rotation, mirroring, and translation of the part. This step sets orientation of the transformation of corrected-image coordinates to object coordinates. The image is not actually transformed; the measurement tools are transformed to align to the image. This reduces the errors in measuring the object.
Creating a Point Set
- Defining a measurement tool. A measurement tool is a collection of line segments, which are used to detect edge points of the object. Each line segment detects one point. A complete measurement tool collects a set of points, which are used to perform measurements.
- Collecting points A line segment from a measurement tool determines a point by sampling the image, and finding the location of an edge in the image. The edge is determined by the point where the image pixel values cross a threshold determined by a percentage of the minimum and maximum values of the pixels along the line.
- Extending point sets to the end points The point set detected by a measurement tool, is extended by following a line or circular arc, determined by the last three points in the point set, until no edge is detected. This is done using a provided step size, and threshold. A binary search is performed to reach the desired accuracy dictated by the input parameters.
Intensity / Density Measurement
Spatial Metric tools
Once a point set is determined by a measurement tool, that point set can be measured in several ways.
- Linear Dimensions
- Circular Dimensions
- Angular Dimensions
- Between Lines
- Circular Arc