Sizing and Roughness Measurement using Laser and Machine Vision

This short document is a description of the on-going projects on sizing and roughness evaluation in the Agricultural Engineering Department at Purdue University. (C.J.Precetti September 24, 1994)

Background

1. Melon detection for robotic harvesting (Miles, 1991)
2. Scratch detection on sanded wood panels (Krutz, Gibson, Precetti, 1991)
3. Seed corn sizing (Krutz, Precetti, 1994)
4. Surface roughness

In these projects a laser line was used to contour the objects in the camera field of view. Image processing was then used to identify the feature of interest: curved object (melon), gap in laser line (groove in wood), average elevation of the line (seed corn height), and line profile (surface roughness). Hardware setup and image processing software are dependent on the application and the analysed feature.

3D seed sizing

This on-going research consists on building and evaluating a 3D seed sizing system. The objective is to calibrate seeds and deliver uniform size seeds. The following figure present a profile of a corn seed acquired with our in-house system:

Sizing is calibrated for the 1/64th inch. The main dimension of seeds can be estimated from the mesh of points acquired from the range image:

Roughness

Surface roughness measures varitions of the surface level. Position of a laser line in an image indicates the height of a surface. Changes in position are recorded. A steady change shows a slope, whereas breaks in the surface level indicate roughness. The current system used in these projects is setup for detection of 1/64 inch. The following image is an example of a change in slope in a styrofoam packaging element.