Examinations used in image processing for screw fastening automation include appearance inspection for screw parts (for foreign substances, burrs, deformations, etc.) and state inspection after fastening.More advanced technology for inspections using image processing as a means to avoid variations in manual inspections is needed in an attempt to seek for a solution with zero defects (with no failures).

(1)Necessity of inspection with image processing

Since human visual inspection has the following limitations, inspections using image processing are needed all the more.

1. Examinations are apt to be affected by the inspector's intuitive judgment, which are hard to be free from subjective interpretations.
2. Standards of judgment may vary depending on the fatigue of the inspector.
3. Long hours of visual inspections may be a limiting factor in the labor environment.
4. Proficiency, which varies among inspectors, may produce difference in inspection productivity. This may affect the labor cost a great deal.

(2)Difficulty in automation of image inspection and various techniques

・Appearance inspection covers a wide range from simple detection of foreign substances to qualitative and intuitive judgment as to contamination, etc. Appropriate image detection sensors and inspecting and identifying algorithms/programs are required based on specific items for testing.

・For these reasons, automation of appearance inspection is often shunned from the viewpoint of Return On Investment.

・For plain recognitions of presence of parts, rough profiles of the shapes, or orientation/direction among a variety of image inspections, a simple and low cost inspection system can be implemented by defining the object under inspection.

・Image sensors include a (1) simple photoelectric sensor, (2) liner image sensor (Fig. 1), (3) matrix image sensor, (4) pattern recognizer (Fig. 2). For applications of (3) and (4), various numerical calculation operations are introduced for pattern recognitions.

・Combination of some photoelectric sensors (phototransistor, photoelectric sensor, etc.) will make determination of rough profile, orientation, and judgment of front and back sides for the object under inspection possible (Fig. 3).

・For applications with the space restriction, the technique of transmitting the image to the image sensor using the optical fiber cable can be used.

・For recognition of small objects, the pictures taken with a digital camera could be displayed on the TV monitor screen to perform image processing by photoelectric sensor using the magnified image.

・Liner image sensors are often used for the case where objects under inspection are moving on the conveyor.Matrix image sensors, composed of photoelectric elements arranged in a matrix form, are applied in the inspection on the two dimensional plane.

・For the application of image sensors, some advanced technique is required. Countermeasures, for example, for environmental fluctuations such as dirt or grime onto the surface, or equalization adjustment of ambient light are necessary. Noise from the accidental reflected light can be avoided by selecting the black color for peripheral equipment around the sensor.

・Equipment processing the image data is responsible for determining the binarization signals (counts of pulse width, etc.), and counting the numerical values ("1" and "0") in each signal area.With the pattern recognition, the data of the area, its characteristics (barycentric position, number of holes, positions, etc.), and lengths will be made available by calculation operations to be used for the processes of comparison or determination.

(3)Handling mechanism and image processing

・Handling mechanism is responsible for rejecting abnormal parts, etc. based on the result (data) from image inspection.