R. Bras. Fisiol. Veg. 4(1): 17-20, 1992 MEASUREMENT OF LEAF AREA WITH A HAND-SCANNER LINKED TO A MICROCOMPUTERl LINDA STYER CALDAS2, CARLOS BRAVO3, HOMERO PICCOLO3 and CARMEN RACHEL S. M. FARIA2 Depto de Botânica, Instituto de Ciências Biológicas, Universidade de Brasilia, CP 04631, Brasília, DF, CEP 70919, Brazil. ABSTRACT- A method was developed for measuring leaf areas with Termos adicionais para indexação: análise de crescimento, a high degree of precision and ease of operation, using an inexpensive herbivoria, fotossíntese, planímetro. hand scanner coupled to an IBM-PC compatible microcomputer. The software program, Area II, was developed by one of us (CB) and is based on pixel counts of the images generated by the scanner, with INTRODUCTION an appropriate calibration factor. A set of 18 leaves of two species of Eucalyptus was photocopied and the areas determined by planimeter, Leaf area must be determined for many physiological studies, gravimetrically (weighing the paper copy of the leaf and comparing the such as growth analysis, photosynthesis and transpiration mea- weight with known areas of the same paper) and with a leaf area meter, surements. Reviews of the techniques normally used for the as well as the scanner. Each leaf was measured four times by each determination of leaf area (e.g. Kvêt & Marshall, 1971; Coombs method and the coefficients of variation of the areas were calculated et al, 1985) cite the leaf area meter as the usual method, although for each method. The coefficients of variation for the leaves of the two planimeters, photogravimetric methods and area-length regres- species of Eucalyptus were of the same order of magnitude as those obtained with the leaf area meter, and smaller than those obtained with sions are also mentioned by these authors. the gravimetric method and the planimeter. Extremely small leaf areas A comparison of these methods for the determination of a large (less than 0.15 cm2) can be measured by using the high resolution (400 number of leaf areas shows that the area-length regressions must dpi) adjustment on the scanner. The method is useful for growth analysis, be performed for each species, and even for different treatments photosynthesis measurements and studies of herbivory, since the within a certain species. For this regression, the area must be damaged areas can be easily completed using the software functions independently determined, using one of the other methods. after the image has been generated. Although the gravimetric method based on the weight of the paper Additional index terms: Growth analysis, herbivory, photosynthesis, cut-out of the leaf tracing (or photogravimetric method, planimeter. when photocopies of the leaves are used), compared to the weight of known areas on the same paper, provides an accurate MEDIÇÕES DE ÁREA FOLIAR COM UM “HAND-SCANNER” measurement of the area, it is a laborious technique when applied ACOPLADO A MICROCOMPUTADOR to a large number of leaves. The planimeter offers a less time-consuming technique, but the RESUMO- Um método foi desenvolvido para medir área foliar com precision is limited, especially for relatively small leaves. um elevado grau de precisão e facilidade de uso, com base num “hand Leaf area meters have been widely adopted for these mea- scanner” de baixo custo acoplado a um microcomputador compatível surements because of their ease of operation. Experience has com o padrão IBM-PC. O programa, Area II, foi desenvolvido por um shown that the other methods are tedious, particularly when a dos autores (CB) e funciona, com um fator de calibração apropriado, na base da contagem dos pixels pretos em imagens geradas pelo large number of leaves must be measured. “scanner”. Foram feitas fotocópias de um conjunto de 18 folhas de duas More recently, methods involving image processing based on espécies de Eucalyptus e as áreas medidas por planimetro, pelo peso video camera images and computer programs for analysis of da fotocópia da folha recortada e por um medidor de área foliar, além these images, have been proposed, for example, Hargrove & do “scanner”. Para cada método, a área de cada folha foi medida quatro Crossley, 1988; Hagerup et al, 1990). These methods permit vezes e os coeficientes de variação das medidas calculadas. Os automatic calculation of leaf areas, and of areas lost to herbivores coeficientes de variação para as folhas das duas espécies de Euca- or disease, depending on the computer programs used. lyptus foram da mesma ordem de magnitude com o “scanner” e com o medidor de área foliar e inferiores aos valores obtidos pelos métodos The following method is presented as a less expensive option gravime’trico e planimétrico. Folhas extremamente pequenas (menores to the leaf area meter and the video camera image processing. que 0,15 cm2) podem ser medidas utilizando o ajuste de alta densidade The alternative method uses a hand scanner linked to an IBM- (400 dpi) do “scanner”. 0 método é útil para análise de crescimento, PC compatible microcomputer. The image of the leaf generated medidas de fotossíntese e transpiração e de herbivoria, pois as áreas by the scanner is transformed to area by an appropriate software danificadas podem ser facilmente completadas com as opções do program, which is based on the pixel counts of the image. software após a geração da imagem. MATERIAL AND METHODS 1 Recebido em 20/03/92 e aceito em 04/06/92. 2 Prof. Adjunto Description of the Scanner Method 3 Aluno de graduação, Depto de Ciências da Computação, Universidade de Brasília, CP 4640, Brasília, DF, CEP 70919. The computer program, Area II, was developed by one of us 4 Prof. Assistente, Depto de Ciências da Computação, Universidade de Brasília, (CB) and is based on a pixel count of the image generated by CP 4640, Brasília, DF, CEP 70919. a hand scanner. A GeniScan TM GS-4500 hand scanner was used 5 Prof. Auxiliar Visitante, Depto de Botânica, CP 04631, Brasília, DF, CEP 70919. R. Bras. Fisiol. Veg. 4(1):17-20, 1992. 18 in the present tests, with ScanEdit II TM software. The scanner uniformly black with an appropriate adjustment of the light-dark has settings for letters (two-tone or black-and-white) or photo- setting of the scanner during image generation. graphs (32 levels of “gray scale”), 100 to 400 dpi resolution and Initial tests before saving the image allowed the adjustment of a variable light-dark adjustment to control the intensity of color the setting. Eucalyptus torelliana leaves, for instance, gave sharply in the image generated. The maximum image width is 105 mm contrasted, homogeneous images from both the dark green, for a single scan but images up to 300 mm wide can be incor- adaxial surface and the lighter green, abaxial surface. Yellow, porated with the merge option. The equipment can be used with senescent leaves of Catharanthus gave very satisfactory images an IBM-PC compatible microcomputer with a minimum of 640 kB against the same white paper background, with a darker adjust- memory and a CGA video monitor. ment. Detached fresh leaves with petioles removed were flattened Necrotic lesions of leaves of Catharanthus and leaflets of woody against a white or light-colored background and covered with a leguminous species ( Dalbergia miscolobium and Pterodon sheet of stiff, transparent plastic before passing the hand scanner pubescens) appeared as white areas in the images generated with appropriate adjustment of the “Light-Dark” control. The by the scanner. Video camera images were analyzed by Lindow scanner was always used with the letter (black-and-white) setting & Webb (1983) to detect the area affected by necrosis, but their and 100 dpi for larger leaves (above O.l5 cm2). Very small leaves system required a black velvet back-ground and red light illumi- (less than 0.15 cm2) from tissue culture potato plants, courtesy nation to maximize contrast. The yellow-green, built-in light source of Lu’cio Fla’vio de Alencar Figueiredo and Eliseu Soares Figueira of the hand scanner gave sufficient contrast for several types of Filho (CENARGEN-EMBRAPA, Brasi’lia, DF), were measured with necrotic lesions to be distinguished in the present, simple system. 200 and 400 dpi settings as well. In the case of photocopies, a sheet of millimeter scaled graph Images were also generated from photocopies of leaves when paper was copied four times to detect possible errors in area a large number of measurements were made. Tracings of the due to reduction or amplification of the leaves by the copier; an outline of leaves which could not be detached are also suitable increase of 1% in the area of the image occurred with the copying for use with this program. In these cases, the image generated machine we used. by the scanner is composed of the outline of the leaf and the The tracings of small leaves, or photocopies with only the “Fill” function in the Graphic Tools of the software can be used border of the leaf appearing, can easily be filled with the “Fill” to easily fill the image with solid black color before saving and function of the scanner software. With the leaf tracings or pho- analyzing the area. tocopies which require filling, it may be necessary to complete The Area II program analyzes the image saved with default the borderline of the leaf before filling to avoid the “overflow” of extension, .TIF, of ScanEdit II. Each leaf is saved as a single, sub- the fill pattern to the area through a break in the outline. This is image file since Area II will provide the total area of all black most easily done by retouching the outline before scanning, in images in the file. The Area II program offers the option of analyzing the points where the line is faint or broken. Nolting & Edwards a series of files and transferring the areas to another file (TEST.DOC, (1985) used a light pen with their video camera system to com- for example). All areas read by Area II will be appended to the plete the margins of partially defoliated leaves prior to estimation TEST.DOC file when this command is used, without erasing of total. undamaged leaf area. previous registers. The images of larger leaves do not fit on the CGA screen used The program was calibrated by generating images with a series in the present system, which makes filling a more laborious task. of squares and rectangles cut from graph paper with millimeter The image must be called up in several parts for filling before divisions. Areas from 0.01 to 108 cm2 were used for the calibration saving for area analysis. If possible, these leaves, or the photo- with at least five independent images for each area. Blank images copied images, can be cut into smaller pieces and the area always gave an area of 0.00 cm2. obtained as the sum of the partial images. This method is pref- erable to the use of the “Merge” function of the scanner software. Comparison with Other Methods Multiple images can be saved in a single file, when the leaf is undulating and must be cut into several pieces to lie flat for Sets of nine leaves of Eucalyptus citriodora and nine leaves scanning. The program will give the total area of all images in of E. torelliana were photocopied and used in all the subsequent a file, based on a total black pixel count. This type of analysis, measurements. One copy was used for the planimeter measure- although it involves saving each leaf as a single file, makes the ments, which were repeated four times on each leaf image with calculation of the leaf areas very rapid (approximately 1 second an Allbrit planimeter. for leaf areas of 50 cm2), when compared with the method of Four copies were used for cutting out the leaf images and Hagerup et al (1990), in which the images of several leaf discs weighing them on a Mettler H35 analytic balance, with a sensitivity are stored in a single file and the algorithm must discover and of 0.1 mg. Thus four separate weighings on the four copies were calculate the areas of separate leaf discs within the file, or picture obtained for each leaf. For calibration of the weight, squares of (20 seconds to estimate the “initial” area without insect feeding 4 cm2 area were cut from the same sheet of paper and weighed and calculate the percentage of the area lost to feeding for each to provide the specific calibration factor for each copy. disk). One copy of each leaf, previously used for the gravimetric In the present method, the “Fill” function of the ScanEdit soft- determination, was measured four times in the Hayashi Denkoh ware was used to complete areas removed by insect herbivory, Model AAM-7 Leaf Area Meter, after calibration of the instrument affected by disease, or removed for measurements of enzyme following manufacturer’s instructions. activity. This was done after saving the image of the original, One copy of the set of leaves was used for scanner measure- damaged leaf. A second image was generated and the difference ments. Four images were generated for each leaf and the areas between the two areas used to calculate the percentage of leaf determined with the Area II program. area affected. This procedure is made easier by retouching the The average area and coefficient of variation for each leaf were presumed outline of the undamaged leaf on a photocopy before calculated for each method. the second scan, when the area affected by grazing is on the edge of the leaf. RESULTS AND DISCUSSION Alternatively, the “Invert Image” command of the ScanEdit software can be used to transform the white areas of damage The images produced from fresh leaves were sharp and to black areas, for direct measurement of the damage, while the R. Bras. Fisiol. Veg. 4(1):17-20, 1992. 19 rest of the leaf area becomes white. The calibration factor was calculated from a set of squares and With the methods which use video cameras to generate the rectangles of known area. Least-squares fit of the known areas images, a separate calibration must be carried out for each distance with the areas determined by the Area II program gave a cor- from the camera to the leaf (Lindow & Webb, 1983: Noting & relation coefficient R2=1.0000; the equation relating the scanned Edwards, 1985; Hargrove & Crossley, 1968). Since this distance area to the known area was: is constant with the hand scanner, a single calibration factor can be used. Scanned Area = 1.001*(Known Area) - 0.0621 The software program Area II was adjusted to give a resolution TABLE 1- Leaf areas measured by four methods. The same leaves were measured four times each by each method. Leaf Planimeter Gravimetric Leaf Area Meter Scanner Eucalyptus Area cv Area cv Area cv Area cv citriodora cm2 % cm2 % cm2 % cm2 % 1 4.5 9.07 4.50 2.04 4.77 0.43 4.46 0.13 2 5.9 12.77 6.55 3.10 6.94 0.82 6.49 0.52 3 6.9 10.91 7.61 1.73 7.79 2.63 7.51 1.20 4 15.3 3.28 15.95 1.37 16.24 0.18 15.30 0.75 5 16.6 2.88 17.50 1.50 18.08 0.36 17.06 0.36 6 21.6 7.41 22.13 1.06 22.62 0.28 21.22 0.43 7 20.8 2.41 23.37 2.18 23.68 0.13 22.56 0.29 8 23.4 3.21 24.BB 2.16 25.20 0.32 23.72 0.27 9 24.8 1.17 25.32 3,17 25.89 0.37 24.82 0.23 Eucalyptus torelliana 1 3.0 13.61 3.70 5.85 4.0O O.94 3.7l 1.33 2 3.3 8.88 4.09 3.35 4.29 0.77 3.97 1.72 3 7.5 5.44 8.34 2.26 8.84 0.72 8.31 1.00 4 7.6 6.28 9.77 1.30 10.29 0.63 9.67 0.45 5 14.5 2.82 15.05 1.51 15.70 O.38 14.89 O.26 6 37.4 4.00 37.36 0.77 38.49 0.20 36.51 0.23 7 42.3 2.82 43.50 2.08 44.82 O.05 42.78 0.23 8 49.5 1.43 49.69 1.66 50.90 0.22 48.64 0.17 9 49.4 0.00 49.93 0.00 51.67 0.00 49.43 0.00 of 0.01 cm2. There is no difference in the area given by the pixels. The Area II program will calculate the area based on the software program for a single image, when the area is determined calibration factor for 100 dpi. If the image was generated with several times. Since there is no need to trace the outline of the the 400 dpi setting, for instance, the given area must be divided leaves or cut the paper form, a high degree of reproducibility is by 16 to correct for the larger number of pixels. The coefficient also possible with the scanner. This can be seen in Table 1, where of variation for areas which appeared as 0.01 cm2 when analyzed the coefficients of variation of the measurements of leaves of at 100 dpi was 77% (Table 2). The same areas, measured at 400 Eucalyptus citriodora and E. torelliana using the scanner were dpi, gave average areas of 0.014 and 0.015 cm2, with coefficients comparable to those obtained with the leaf area meter. The of variation of 4.4 and 5.51% (Table 2). Areas in the range of O.1O coefficients of variation with the scanner were one-half to one- to 0.25 cm2 can be measured at 400 dpi with a coefficient of tenth of those obtained with the gravimetric method and one-third variation ranging from 0.5% to 2.3% (Table 2) while even smaller to one-fiftieth of the coefficients obtained with the planimeter. areas (from 0.015 cm2) were measured with a maximum coef- The area analysis system of Nolting and Edwards (1985) gave ficient of variation of 11.6% (Table 2). significantly lower standard deviations (P<0.05) for repeated measurements of the same objects than the leaf area meter. TABLE 2- Minimal leaf areas measured at three resolutions. The same leaves were The average leaf areas obtained with the scanner for the smaller seasured four times each at each resolution. leaves are closer to the values from the gravimetric method than 100 dpi 200 dpi 400 dpi to the values from the planimeter or the leaf area meter. Linear regression of the measurements of the different methods as a Area cv Area cv Area cv function of the scanner areas showed the closest agreement Leaf cm2 % cm2 % cm2 % between the gravimetric method and the scanner (slope of 1.015), 1 0.01 77 0.012 20.2 0.014 4.4 while the planimeter and the leaf area meter gave slightly higher 2 0.01 77 0.009 28.6 0.015 5.5 values for the areas (slopes of 1.020 and 1.028, respectively). 3 O.02 38 0.013 36.1 0.016 6.8 4 O.02 29 0.022 19.5 O.022 11.6 Lindow & Webb (1983) showed a good correlation (coefficient 5 O.03 33 0.023 14.0 0.027 2.2 of determination = 0.99) between the areas of bean leaves 6 0.03 50 0.028 8.9 0.032 9.0 measured with their digitized video image analysis and the areas 7 0.02 29 0.032 3.9 0.036 2.3 measured with a planimeter. They did not describe the calibration 8 0.03 31 0.035 10.1 0.041 4.0 procedure for the image analysis and used the planimeter 9 0.02 22 0.036 8.0 0,041 3.1 measurements as the true area. 10 0.06 22 0.068 6.8 0.073 1.9 11 0.10 8 0.108 3.3 0.111 0.5 A high degree of precision can be achieved in measuring very 12 0.11 7 0.118 2.0 0.126 0.5 small leaf areas (less than 0.l5 cm2) by adjusting the scanner 13 0.12 8 0.133 4.6 0.145 1.2 to 200 or 400 dpi, instead of 100 dpi (Table 2). An area of 0.01 14 0.15 4 0.155 0.0 0.157 2.3 cm2 appears as 16 pixels after scanning at 100 dpi; at 200 dpi, 15 0.22 7 0.233 2.6 0.256 0.9 the same area gives an image with 64 pixels and, at 400 dpi, 256 16 0.24 3 0.243 4.6 0.264 1.3 R. Bras. Fisiol. Veg. 4(1):17-20, 1992. 20 In comparison, the smallest area measured with the video Copies of the program are available from the authors. camera/digitizer system of Lindow & Webb (1983) was 20 cm, while Hargrove & Crossley (1988) indicated a minimum of 5 cm2 ACKNOWLEDGMENTS and Hagerup et al (1990) measured leaf discs with an area of 1.65 cm2. Nolting & Edwards (1985) were able to measure objects We thank Carlos Lazarini Fonseca, of CPAC/EMBRAPA, for his with a minimum area of 0.25 cm2. The high degree of resolution help with the Leaf Area Meter; Heloisa Sinatora Miranda and Ruy of the present method makes it particularly suitable for the de Araujo Caldas, for critical reading of the manuscript, and Kiniti measurement of small leaf areas; it is more laborious when large Kitayama, for providing useful bibliography. leaves must be measured, frequently entailing a preparation process to cut the leaves up into smaller parts, which will also REFERENCES lie flat, without undulations. In summary, the present method proved to be easy to use, in . COOMBS, J.; HALL, D.O.; LONG, S.P & SCURLOCK, J.M.O. Techniques that the contours of the figure do not need to be followed manually. in bioproductivity and photosynthesis. 2 ed. Oxford, Pergamon Press, It is also extremely rapid (as many as 120 measurements of 1985. pp 45-46. leaves with areas in the range of 5 to 15 cm2 have been made , HAGERUP M.; SONDERGARD, I. & NIELSEN. J. K. Measurements of areas consu- med from leaf discs by chewing phytophagous insects: Description of a in one hour) and less tiring than other, traditional methods. new method involving image processing.Entomologia Experimentalis et Another advantage is that the equipment (the hand scanner, Applicata., 57:105- 113, 1990. excluding the microcomputer) is not only inexpensive. with a cost HARGROVEC, W.W. & CROSSLEY Jr., D.A. Video digitizer for the rapid measure- which may be less than one-tenth of that of more sophisticated, ment of leaf area lost to herbivorous insects. Annals of the Entomological Society of America, 81:593-598, 1988. specialized leaf area meters, but it also can be used in many other KVE‘’T, J & MARSHALL. J.K. Assessment of leaf area and other assimilating plant applications, such as desktop publication. Obviously, the same surfaces. In: Sesta’k. Z.; Catsky’, J. & Jarvis, P.G., eds. Plant Photosynthetic software program and method can be applied to measure the Production. Manual of Methods. The Hague, Dr W. Junk. 1971. pp. 517-555. area in closed figures on graphs or maps, for integrating daily LINDOW, S.E. & WEBB, R.R. Quantification of foliar plant disease symptoms by microcomputer-digitized video image analysis. Phytopathology, 73:520-524, totals of CO2 fixation, transpiration or many other applications. 1983. This flexibility makes the program of general use, although . NOLTING, S.P & Edwards, C.R. Defoliation assessment using video imagery and additional calculations are required for determining areas lost to a microcomputer. Bulletin of the Entomological Society of America, 31(4):38- herbivores or disease. 42, 1985. R. Bras. Fisiol. Veg. 4(1):17-20, 1992.
Pages to are hidden for
"MEASUREMENT OF LEAF AREA WITH A HAND-SCANNER "Please download to view full document