A Laser Optical Method for Detecting Corn Kernel Defects

S. Gunasekaran, M. R. Paulsen, G. C. Shove ASSOC. MEMBER MEMBER MEMBER

ASAE ASAE ASAE

ABSTRACT

AN opto-electronic instrument was developed to examine individual corn kernels and detect various

kernel defects according to reflectance differences. A low power helium-neon (He-Ne) laser (632.8 nm, red light) was used as the light source in the instrument. Reflectance from good and defective parts of corn kernel surfaces differed by approximately 40%. Broken, chipped, and starch-cracked kernels were detected with nearly 100% accuracy; while surface-split kernels were detected with about 80% accuracy.

INTRODUCTION

Corn is one of the most important and widely produced grain crops in the U.S. Hence, maintaining high quality of corn is very important. Quality of corn generally refers to plumpness, soundness, and cleanliness which are reflected in the test weight, moisture, damage, and broken corn and foreign material (BCFM) (USDA, 1978). Quality problems with corn have increased as greater amounts and proportions of corn is combine shelled, artificially dried, and moved through market channels (Anderson, 1972). Combine shelling and artificial drying can induce stress cracks in corn kernels (Thompson and Foster, 1963; Roberts, 1972). Stress cracks are very fine fissures in kernel endosperm measuring about 53 pm in width (Gunasekaran et al., 1985a). Stress-cracked and externally damaged kernels break up more readily during transportation and handling and give rise to large amounts of BCFM (Paulsen and Hill, 1977). Corn damaged during these processes will yield lower percentages of large grits in dry milling, and starch in wet milling than undamaged corn. Large grits and starch are two valuable prime products of dry and wet milling, respectively. Also, damaged corn is more readily attacked by microorganisms; and has a reduced storage life. High amounts of BCFM greatly increase the risk of dust explosions in grain elevators; restrict airflow while

Article was submitted for publication in April, 1985; reviewed and approved for publication by the Electric Power and Processing Div. of ASAE in November, 1985.

This study was a part of Project No. 10-0359 of the Agricultural Experiment Station, College of Agriculture, University of Illinois at Urbana-Champaign.

The authors are: S. GUNASEKARAN, Assistant Professor, Agricultural Engineering Dept., University of Delaware, Newark; and M. R. PAULSEN, Associate Professor, and G. C. SHOVE, Professor, Agricultural Engineering Dept., University of Illinois, Urbana.

Acknowledgments: Appreciation is expressed to Dr. N. S. Vlachos, and Dr. L. D. Savage, Jr., Mechanical Engineering Dept., University of Illinois, for their helpful suggestions in designing the instrument.

drying; and contribute to non-uniform flow of materials during processing. Thus, evaluating corn for quality characteristics is very important.

Present corn quality evaluation methods are largely manual. These methods are time consuming, laborious and often give imprecise results. To obtain an easy, fast and more accurate evaluation of corn quality an improved, automatic method is necessary. Potential methods of automatic corn quality evaluation are discussed in Gunasekaran and Paulsen (1984).

Optical methods have been successfully used to detect defects in biological materials. Gunasekaran et al. (1985b) presented a descriptive account of appropriate optical methods for nondestructive quality evaluation of agricultural and biological materials. Investigations on corn include determination of damage in a large sample of corn (Johnson, 1962) and detection of mold contamination (Birth and Johnson, 1970) which indicate that defects consistently affect the optical properties of corn kernels. This paper discusses the design and development of an optical instrument to detect various kernel defects based on light reflectance measurements using a low power helium-neon (He-Ne) laser as the light source.

OBJECTIVES

The objectives of this investigation were: 1. To establish a criterion to identify various corn

kernel defects. 2. To design and develop an instrument to detect

kernel defects based on the established criterion. 3. To examine corn kernel surfaces with various

defects and identify the defect categories.

THEORETICAL CONSIDERATIONS

When light is incident on any material it may be reflected, transmitted, and absorbed. The degree to which these phenomena take place depends on the nature of the material; particular wavelength of the electromagnetic spectrum used; and the angle of incidence.

Based on its optical properties, an object may be transparent, translucent, or opaque. In general, agricultural products are opaque, although most transmit light at certain wavelengths (Birth and Zachariah, 1973). It was commonly believed that the colors of natural objects are seen by means of light reflected from the surface, known as regular reflectance. Birth (1976) recognized that light must be transmitted through the pigment within cells to produce a colored appearance. Light transmitted in this manner encounters randomly oriented internal interfaces and is

294 © 1986 American Society of Agricultural Engineers 0001-2351/86/2901-294$02.00 TRANSACTIONS of the ASAE

Fig. 1—Relative positions of photo detector for measuring predominantly the body reflectance! 1), and the regular reflectance(2).

reflected back. A fraction of this reflected radiation is transmitted back through the initial interface and is named body reflectance. The body reflectance is almost always diffuse and is more significant than regular reflectance for quality evaluation of food products.

Most optical equipment for quality evaluation of biological materials measures total reflectance, the sum of regular and body reflectance. Information regarding the kernel defects is primarily contained in the body reflectance. When regular reflectance predominates the light measurement, any information contained in the body reflectance may be masked. Hence it is important to design a system in which the regular reflectance does not predominate. Fig. 1 illustrates how regular and body reflectance are involved in measurement.

The success of an optical measurement system largely depends, among others, on the quality of the light used for the measurements. With its highly intense, coherent, directional, and well defined diameter a laser beam is unsurpassed in its quality as a monochromatic light source. Moreover, the laser beam is more readily adaptable for optical uses than other light sources. Corn kernel defects range from a few microns, for stress cracks, to several millimeters, for chipped and broken kernels. To indicate defects of these dimensions the light beam should be sensitive to the smallest defect dimension. The equation applicable to focusing a laser beam is:

4Xf

where: s = diameter of focused spot, yim A = wavelength of laser light, ^m f = focal length of the focusing lens, mm d = diameter of the laser beam, mm

THE INSTRUMENT

Principle of Operation Optical properties of a good corn kernel surface are

consistently affected by the presence of various defects. This is the basis for developing an opto-electronic instrument for detecting various corn kernel defects.

The laser beam light was focused to about 18 jum in diameter. The kernels, mounted on a sliding plate, were traversed across the stationary laser beam. As the kernel surface was scanned, the light reflected from the surface was collected on a photomultiplier tube. The

%4«/T

Fig. 2—Laser optical instrument for corn kernel defect detection. 1-He-Ne laser; 2-Optical bench; 3-Photomultiplier tube; 4-DoubIe convex lens; 5-Beamsplitter; 6-Microscope objective; 7-Kernel drive assembly.

Fig. 3—Kernel scanning section of the laser optical instrument. 1-Photomultiplier tube; 2-Double convex lens; 3-Beamsplitter; 4-Microscope objective; 5-Corn kernels.

photomultiplier tube converted the intensity of light falling on it into a proportional electrical signal indicating the type of defect being examined.

Fig. 2 is a photograph of the instrument; and a close-up view of the kernel scanning section is presented in Fig. 3. A brief description of the major components of the instrument and their function is given below:

Laser A helium-neon (Ealing*, Model 25-0803) was used as

the light source. The laser operated at 632.8 nm wavelength (red light) with a linearly polarized, continuous wave output of 1 mW. The diameter and divergence of the laser beam are 0.63 mm and 1.3 mm radians, respectively.

Photomultiplier tube (PMT) Hamamatsu, Model R928 photomultiplier tube with

an extended red, high sensitivity, multi-alkali

T r a d e names are used in this publication solely for the purpose of providing specific information. Mention of a trade name, proprietary product or specific equipment or company does not constitute a guarantee or warranty of the University of Illinois and does not imply approval of the named product to the exclusion of other products that may be suitable.

Vol. 29(1):January-February, 1986 295

< E

'•5 to

CO

500 600 700 800 1000 1200

Wavelength , nm

Fig. 4—Spectral response characteristics of the Hamamatsu, R928 photomultiplier tube.

definition at the wavelength of He-Ne laser (632.8 nm). The lens surfaces were coated with a special single layer, broadband anti-reflectance coating for minimum reflectance from the lens surfaces.

The focal length of the objective was 14 mm and the numerical aperture was 0.17. The objective lens performed two functions in the instrument by focusing the laser beam to a small spot size, and by collecting the light reflected from the kernels. At the focal point, the light beam was 18 jum in diameter according to equation [1]. Since the lens was placed at a distance equal to its focal length from the kernels, it very effectively collected the reflected light from the kernels eliminating need of an additional lens for this purpose; and also provided a parallel, collimated beam of collected light back toward the beamsplitter for signal measurement. This arrangement along with the beamsplitter was compact and convenient.

photocathode sensing element was used. This tube was sensitive to electromagnetic radiation in the range of 185 to 930 nm with a peak response at 400 nm. The spectral response of this photomultiplier tube is shown in Fig. 4.

The PMT converts the intensity of light falling on its window into a proportional electrical signal. Special circuitry was developed to power (1000 volts DC at 2 mA) the PMT, and amplify the signal for recording. To avoid any stray light interference, the window of the PMT was covered with a black sheet leaving an opening of about 2 x 2 mm square to collect the signal light.

Beamsplitter A Corion, Model BS550, 50/50-beamsplitter was

positioned at a 45 deg angle to the laser beam line to split the beam into two equal and perpendicular halves. One half of the beam continued in the original direction of the beam; the other toward the base of the instrument and was not used. The same action was repeated in the opposite direction with the reflected light collected from the kernel surface by the microscopic objective. However, this time the beam emerging perpendicular to the line of laser was used for measurements. The action of the beamsplitter as applied in the instrument is depicted in Fig. 5. The beamsplitter reduced the intensity of light collected by the PMT by about 75%. However the amount of light collected was found to be sufficient for defect measurements.

Microscopic Objective Lens A high precision Ealing, Model 24-8690 achromatic

objective lens was used. This objective lens was designed and adjusted to an infinity focus to provide the best

I To phototube

From laser

Not used

To Kernel

Reflected from Kernel

Not used Beamsplitter

Double Convex Lens A double convex lens of 5 cm focal length was

positioned over the beamsplitter to focus the light collected from the corn kernels onto the PMT.

Kernel Drive Assembly The kernel drive assembly consisted of a 25 x 10 x 10

cm metal frame supported by four 45 cm adjustable legs. The frame was provided with a sliding plate through a threaded screw and gear assembly. A constant speed, reversible motor was used to operate the unit. The instrument also had a 1.15 m long optical bench (Central Scientific, Model 85808), which provided stable support for the entire instrument and lens holders.

EXPERIMENTS

Corn kernels of four genotypes (FR4A x FR4C) x Mol7, FRB73 x Mol7, FRB73 x FR16, and Mol7 x HlOO, that are widely grown in the U. S. corn-belt were used. The FRB73 x Mol7 is more popular than the rest; and Mol7 x HlOO is large seeded and contains large proportions of floury endosperm. Relative values of breakage susceptibility for (FR4A x FR4C) x Mol7, FRB73 x Mol7, Mol7 x HlOO, and FRB73 x FR16 genotypes are the lowest to the highest, respectively (Paulsen et al., 1983). These kernels were obtained from high-temperature dried corn samples. Stress-cracked kernels from these samples were grouped separately. Other defect categories such as broken (loss of original shape), chipped (loss of a part of pericarp and vitreous endosperm exposing white floury endosperm but retaining the kernel shape), starch-cracked (open pericarp and exposed endosperm), and surface-split (open pericarp and exposed endosperm) kernels were

Fig. 5—Laser beam path through the beamsplitter. Fig. 6—Corn kernels mounted onto a sample holder plate to scan the kernel surfaces for defect detection.

296 TRANSACTIONS of the ASAE

o > CO

2 2 CO

8h

J ^ - ^ A K J 1 ^ ^

\- f

0 2 4 6 8

Kernel width, mm

Fig. 7—Photomultiplier signal output from a good kernel surface traversed perpendicular to the laser beam.

obtained by running the kernels through a centrifugal impactor.

Kernels to be examined for defects were glued (Elmer's Glue-all adhesive) onto a plastic sample holder plate (Fig. 6). The kernel tip cap was slightly sanded prior to mounting on the plates. Several sample holder plates were prepared to be used individually in the drive mechanism of the instrument to scan the kernel surfaces. The plate was positioned in front of the microscopic objective at a distance of 14 mm (focal length of the microscopic objective). The kernels were traversed across the stationary laser beam at a speed of approximately 2.25 cm/min. The voltage signals from the PMT were plotted with an x-y plotter as the kernel surfaces were scanned.

The kernels were scanned with their surfaces subtending various travel angles of 90, 85, 80, 75, and 70 deg. Travel angle was changed by appropriately positioning the drive mechanism unit while maintaining the 14 mm distance between the kernel surface and the microscopic objective. In all experiments, the kernels were scanned on the side opposite the germ at about midway between the crown and tip cap of the kernel.

RESULTS AND DISCUSSION

Typical signal recordingt when a good corn kernel was scanned normal to the laser beam is shown in Fig. 7. The signal was high when the laser beam hit the edge of the kernel and remained high until the entire kernel surface was scanned. However, the signal fluctuated significantly during the kernel surface examination presumably due to the inhomogeneities on the kernel surface. The signal fluctuations interfered with the change in the signal level due to cracks and splits. This problem was overcome by traversing the kernels at angles other than 90 deg to the laser beam. Fig. 8 represents the signal recording when the same good kernel was scanned at different travel angles. At travel angles of 80 and 85 deg the signal fluctuations were less. This was due to the fact that at angles other than 90 deg, the reflected light collected by the microscopic objective contained more body

tAll signal recording figures are for (FR4A x FR4C) x Mol 7 kernels, signals from kernels of other genotype are similar.

> 0

09

c 2 CO

i-^OJtjw

\- f 9 0 85°

r 6 8

p^^^^v^n/y.

8 0 75 v

0 2 4 6 8 0 2 4 6 8-

Kernel width, mm

Fig. 8—Effect of travel angle on photomultiplier signal output from a good corn kernel surface.

reflectance than regular reflectance. The signal, however, was attenuated at angles of 75 deg or less. From testing over 100 kernels, a travel angle of 80 deg to the laser beam was determined to be the best for detecting different defect categories.

Fig. 9 is typical of signal recording when chipped kernels were examined. The peak portions of the signal correspond to the defective parts of the kernel showing the white inner endosperm. Signals from starch-cracked and surface-split kernels are shown in Fig. 10. From these figures, the type of defect as well as the extent of

10

~ 6 O >

55 2

fl ^ W L-JYI

0 2 4 6 8 0 2 4 6 8

Kernel width, mm

Fig. 9—Photomultiplier signal output from corn kernels with parts of exposed floury endosperm on the surface (chipped kernels) traversed at an angle of 80 deg to the laser beam.

Vol. 29(1):January-February, 1986 297

6f-

o >

"3 c 3 5

oM.

6f-

100

^ 8 0 r

o « 6 0

CD

cc 4 0

2 0

IT—T Surface split

2 h

o^-v-

iu^k

P I ^ J S J U - J

OMr-

rsv

rv^

Crack

4 6 8 0 2

Kernel width, mm

" 2 - 4 6 0 2 4 6 8

Kernel width, mm

Fig. 10—Photomultiplier signal output from cracked and surface split corn kernels traversed at an angle of 80 deg to the laser beam.

damage on the kernel surface can be determined. The smallest defect detected was in the order of 0.1 mm in width. The signal level, when represented as relative reflectance (Fig. 11), indicated a reflectance difference of 40% between the sound yellow part of the kernel and the exposed white floury endosperm part. Table 1 presents the performance data of the instrument in detecting different corn kernel defects based on the average for 50 kernels in each corn variety and in each defect category.

Fig. 11—Relative reflectance of good and exposed floury endosperm parts of a corn kernel surface.

An overall efficiency of 100% was obtained in detecting broken and chipped kernels; and 96 and 79% for starch-cracked and surface-split kernels, respectively.

Difference in light reflectance alone was not sufficient to detect the stress cracks using the equipment developed because light reflectance is a surface phenomenon whereas the stress cracks are internal defects beneath the pericarp. Also, most of the radiation returned as body reflectance penetrated a relatively short distance into the kernel so the stress cracks did not influence adequately to be sensed.

1.

CONCLUSIONS

Light reflectance difference between good and defective kernel surfaces can be used to determine several types of corn kernel defects.

2. An instrument was designed and developed to detect corn kernel defects based on the light reflectance difference at 632.8 nm (red light).

(continued on page 304)

TABLE 1. PERFORMANCE DATA OF THE LASER OPTICAL INSTRUMENT IN DETECTING DIFFERENT CORN KERNEL DEFECTS (BASED ON AVERAGE FOR 50 KERNELS IN EACH

VARIETY AND IN EACH DEFECT CATEGORY).

Kernel defect

Broken

Chipped

Starch-cracked

Surface-split and minor cracked

Genotype

(FR4AxFR4C)xMol7 FRB73xMol7 FRB73xFR16 Mol7xH100

(FR4AxFR4C)xMol7 FRB73xMol7 FRB73xFR16 Mol7xH100

(FR4AxFR4C)xMol7 FRB73xMol7 FRB73xFR16 Mol7xH100

(FR4AxFR4C)xMol7 FRB73xMol7 FRB73xFR16 Mol7xH100

Signal from non-defective

surface, V

5.78 5.70 5.63 5.55

5.81 5.78 5.60 5.61

5.85 5.71 5.65 5.50

5.75 5.73 5.55 5.58

±0.30 ± 0.25 ±0.30 ± 0.31

± 0.25 ± 0.22 ±0.28 ± 0.21

± 0.28 ±0.20 ±0.25 ± 0.19

± 0.24 ±0.20 ± 0.19 ± 0.18

Signal difference between defective and non-defective

surface, V

4.05 ± 0.45 3.89 ± 0.43 3.93 ±0.50 3.78 ± 0.37

3.94 ± 0.38 3.78 ± 0.46 3.90 ± 0.50 3.85 ± 0.32

3.63 ± 0.65 3.51 ± 0.70 3.47 ± 0.47 3.50 ± 0.81

1.17 ± 0.50 1.02 ± 0.54 0.93 ± 0.48 1.15 ± 0.61

Defect detection efficiency, %

For each variety

100 100 100 100

100 98

100 100

98 94 94 96

84 78 74 80

Overall

100

99.5

95.5

79.0

298 TRANSACTIONS of the ASAE

line routings and/or alternative transmission tower designs. Because certain impacts to agriculture occur over a period of time, determination of the present worth of residual impacts enables a basis of comparison between alternatives.

A last and very real problem is the comparison of residual agricultural impacts with residual impacts on other resources along the transmission line such as visual, archeological, recreation and environmentally protected resources. Final routing decisions for transmission lines must ultimately consider not only the most economic routing but also the one least disruptive to the environment as a whole.

References 1. Anonymous. 1980. APS/SDG&E interconnection project —

draft environmental document. U.S. Department of the Interior, Bureau of Land Management and Public Utility Commission, State of California.

2. Anonymous. 1977. Economic effects of hydro transmission towers on orchard and vineyard operations. Environmental Resources Section, Department of Forestry, Hydraulic Generation and Transmission Division, Ontario Hydro, Canada.

3. Fortin, J. M. and C. Vigneault. 1981. Impact of electric towers on farm machinery operations. ASAE Paper No. 81-3503, ASAE, St. Joseph, MI 49085.

4. Gustafson, Robert J., R. Vance Morey, Vernon R. Erdman, and Edwin R. Hendrickson. 1980. Interaction of center pivot irrigation and electrical transmission. TRANSACTIONS of the ASAE 23(2):485-489.

5. Gustafson, Robert J., Philip D. Grumstrup, Edwin R. Hendrickson, and Merle P. Meyer. 1980. Land lost from production

3. There is a 40% difference between relative light reflectance from good yellow part and the exposed white floury endosperm part of the kernel.

4. Broken and chipped kernels were detected with an accuracy of near 100%; starch-cracked and surface-split kernels were detected with an accuracy of 96 and 79%, respectively.

5. A kernel travel angle of 80 deg to the laser beam was most suitable for detecting various kernel defects. At travel angles of 75 deg and less, the PMT signals were greatly attenuated.

6. Difference in light reflectance was not sufficient to detect stress cracks using the equipment developed.

References 1. Anderson, D. E. 1972. Economic aspects of conserving quality

of export corn. Corn and Soybeans Grain Damage Symposium. Ohio State University, Columbus, OH.

2. Birth, G. S. 1976. How light interacts with foods. In: Quality detection in foods. Gaffney, J. J. (ed.), ASAE, St. Joseph, MI 49085.

3. Birth, G. S., and R. M. Johnson. 1970. Detection of mold contamination in corn by optical measurements. J. Assoc. Off. Anal. Chem. 53(5):931-936.

4. Birth, G. S., and G. L. Zachariah. 1973. Spectrophotometry of

under and around electrical transmission line structures. TRANSACTIONS of the ASAE 23(1): 180-184.

6. Hancock, J. T. and D. R. Kallesen. 1980. Induced current measurements on a haystack wagon under a 345,000 volt line. TRANSACTIONS of the ASAE 23(1):185-188.

7. Lee, J. M. and G. L. Reiner. 1980. Transmission line electric fields and the possible effects on livestock and honeybees. TRANSACTIONS of the ASAE 23(l):281-286.

8. Osborn, Tim C , Robert N. Shulstad, and Alan D. McQueen. 1982. Overhead electric transmission line and support structures: cost and yield effects on the production of cotton and soybeans. Department of Agricultural Economics and Rural Sociology Staff Paper No. 1982-1, University of Arkansas.

9. Roy, W. R., W. C. Pokorny, and T. Vinh. 1981. Induction effects of a 765 KV transmission line on a center pivot irrigator. ASAE Paper No. 81-3504, ASAE, St. Joseph, MI 49085.

10. Roy, W. R. 1981. Studies of field crops near high voltage transmission lines. ASAE Paper No. 81-3604, ASAE, St. Joseph, MI 49085.

11. Stetson, L. E., D. R. Kallesen, K. E. Ewy, and Richard E. Hanson. 1981. Tests evaluating conduction of electrical current through irrigation streams contacting overhead power lines. ASAE Paper No. 81-3505, ASAE, St. Joseph, MI 49085.

12. Anonymous. 1979. The effects of electric transmission lines and tower on agriculture. Prepared for Land Department, Pacific Gas and Electric Company, San Francisco, CA, by Resources International, 402 Rowell Building, Fresna, CA 93721.

13. Anonymous. 1977. The effects of hydro transmission towers on farm operations in western and eastern Ontario — a synthesis of the Ridgetown and Kemptville studies. Environmental Resources Section, Department of Forestry, Hydraulic Generation and Transmission Division, Ontario Hydro, University Avenue, Toronto, Ontario M5G1X6, Canada.

14. Anonymous. 1979. Transmission lines and agricultural land. Property Division, 60 Murray Street, Toronto, Ontario M5G1X6, Canada.

agricultural products. TRANSACTIONS of the ASAE l(3):548-552. 5. Gunasekaran, S., and M. R. Paulsen. 1984. Potential methods

for automatic detection of corn kernel defects. ASAE Paper No. 84-3558, ASAE, St. Joseph, MI 49085.

6. Gunasekaran, S., S. S. Deshpande, M. R. Paulsen, and G. C. Shove. 1985a. Size characterization of stress cracks in corn kernels. TRANSACTIONS of the ASAE 28(5):1668-1672.

7. Gunasekaran, S., M. R. Paulsen, and G. C. Shove. 1985b. Optical methods for nondestructive quality evaluation of agricultural and biological materials. J. Agric. Eng. Res. 32(3):209-241.

8. Johnson, R. M. 1962. Determining damage in yellow corn. Cereal Science Today 7(1): 14-15.

9. Paulsen, M. R., and L. D. Hill. 1977. Corn breakage susceptibility in overseas shipments — two case studies. TRANSACTIONS of the ASAE 20(3):550-557.

10. Paulsen, M. R., L. D. Hill, D. G. White, and G. F. Sprague. 1983. Breakage suscept ibi l i ty of corn-bel t genotypes . TRANSACTIONS of the ASAE 26(6): 1830-1841.

11. Roberts. H. J. 1972. Corn damage and its effect on dry milling. Corn and Soybeans Grain Damage Symposium, Ohio State University, Columbus, OH.

12. Thompson, R. A., and G. H. Foster. 1963. Stress cracks and breakage in artificially dried corn. Marketing Research Bulletin No. 632, TFRD, ARS, USDA.

13. USDA. 1978. The Official United States Standards for Grain. Federal Grain Inspection Service, Inspection Division, U.S. Department of Agriculture.

Laser Detecting Com Defects (continued from page 298)

304 TRANSACTIONS of the ASAE