You are on page 1of 7

International Journal of Emerging Technology and Advanced Engineering

Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012)

Application of image processing in seed technology: A survey


Mrs. Archana Chaugule
Department of IT, Army institute of technology, Pune
archna.isha@gmail.com Abstract This paper aims to present a review of seed technology, seed germination and vigor methods using image processing. Computeraided image analysis techniques have been recently developed in monitoring seed growth and vigor. Their integration with the standard germination test is needed to describe the germination performance of a seed sample with high accuracy. The use of various modern image acquisition techniques combined with image processing techniques have allowed developing automated seed quality tests,. The two main limitations of performing a vigor test manually are 1) results of a vigor test may vary from laboratory to laboratory because of the subjective nature of most vigor tests and 2) many vigor tests take excessive time to acquire results. These two limitations can be addressed by designing computer software that measures the seedlings represented by a digital image and computes the vigor index from those measurements. Several theories of seed germination and vigor are briefly mentioned. Methods are classified into several groups. KeywordsArtificial Vision System, Living Ability Index, Seed Germination, Seed Vigor, Seed Quality Tests, Vigor Index.

I. INTRODUCTION Standard germination tests support the seed vigor testing process. Vigor tests are designed to mimic poor seeding conditions to find out how the seed lot will perform under stress. It is the exact opposite of a germination test, where seed is grown under optimum conditions. A. Basics of Seedling and Seed Technology. Seeds are the beginning and the end of most agricultural practices. The ways in which seeds function, their physiology, biochemistry, molecular biology, and genetics are critically important for agricultural success. Seeds are self-contained units, in contrast to the plants that develop after germination, due to the materials stored in the seeds. Environmental requirements for germination are fewer and simpler than those for whole plant development, so germination is relatively independent of the environment for a considerable period of seedling development. 153

This assumption is based on the observation that a seedling does not photosynthesize; therefore, it requires neither light (except for regulatory or triggering functions) nor CO2 for its proper development until the seedling breaks through the soil surface. Nevertheless, other environmental factors are needed, such as water, temperature, and oxygen. Seed testing is the cornerstone of all other seed technologies. It is the means by which we measure the viability and all the physical factors that regulate the use and maintenance of seeds. Everything that is done with seeds should have some test information to guide the work and ensure high quality. Seed tests tell if a crop of seeds is worth collecting, if handling procedures are correct, and how many potential seedlings are available for regeneration. Seed testing is determining the standards of a seed lot namely physical purity, moisture, germination and thereby enabling the farming community to get quality seeds. In seed evaluation, germination is defined as the emergence and development from the seed embryo of those essential structures which, for the kind of seed in question, indicate its ability to produce a normal plant under favorable conditions. Conditions used for the standard germination test, however, are almost in direct contrast with conditions to which seeds are normally subjected in the field. In the standard germination test, because moisture and temperature during the test are optimum and the germination substrata, unlike soil, are not loaded with microorganisms, herbicides, fertilizers, fungicides or systemic insecticides, weak deteriorated seeds may be capable of producing a normal seedling. In many instances, seed lots of apparently equal quality as indicated by germination percentage will produce largely different responses in field emergence. Therefore, determining "the degree of aliveness" (vigor) of a seed is just as important as determining whether the seed is alive. Clearly, a germination test alone is not enough to assess seed quality, vigor test is also required. Germination test and Seed vigor tests have traditionally been used to determine deterioration of seed samples.

International Journal of Emerging Technology and Advanced Engineering


Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012) Standard germination percentage and the seed vigor of the seed lot can be used to measure seed lot quality. B. Seed Germination Germination is defined as the emergence and development from the seed embryo of those essential structures which, for the seed in question, are indicative of the ability to produce a normal plant under favorable conditions (AOSA, 2000). Germination depends on the quality of the seed lot under observation. Seed lots perform poorly under commercial growing conditions if they do not germinate at high percentages under near-optimal conditions. 1) Germination Test: Seeds can be germinated on various media. Sand, sand and perlite mixtures, potting mixtures, soil, and various papers blue blotters, white blotters, or crepe-cellulose papers . Specifying the medium helps to assure uniformity in test results. At the end of the test period, seedlings can be evaluated and categorized as follows: Normal: Seedlings that possess essential structures those are indicative of their ability to produce useful mature plants under favorable field conditions. Abnormal: Seedlings that exhibit some form of growth but have insufficient plant structures to maintain a healthy plant, such as missing roots or shoots. Fresh Seeds: Seeds that have failed to germinate but have imbibed water. They appear firm, fresh and capable of germination, but remain dormant. Dormant Seeds: Viable seeds (other than hard seeds) that fail to germinate when given the prescribed or recommended germination conditions. Hard Seeds: Seeds that remains hard at the end of the prescribed test period, because their seed coats are impermeable to water. Dead Seeds: Seeds that cannot produce any part of a seedling. C. Seed vigor A vigor test reveals ability of seed lot to withstand a variety of different stress factors. In a vigor test, the seed is introduced to a stressful environment unfavorable to seedling development. This environment can be cool, cold or warm, or a combination of either high humidity and high temperatures, or heavy moisture at low temperatures. If the seed lacks vigor, one or more of these created stressors will suppress seedling growth but, if the seed is vigorous, it will withstand one or all of these stressors and grow as if it were on stimulants. 154 Excerpts of two scientific definitions for vigor from the International Seed Testing Association (ISTA) and the Association of Official Seed Analysts (AOSA) are given below. ISTA: Seed vigor is the sum of those properties which determine the potential level of activity and performance of the seed or seed lot during germination and seedling emergence. AOSA: Seed vigor comprises of those seed properties which determine the potential for rapid, uniform emergence and development of normal seedlings under a wide range of field conditions. 1) Seed Vigor Test: Vigor testing is an important component of seed testing because its more sensitive test than germination, and because loss of vigor may be noted much earlier than loss of germination. Vigor testing does not only measure the percentage of viable seed in a sample, it also reflects the ability of those seeds to produce normal seedlings under less than optimum or adverse growing conditions similar to those which may occur in the field. Seeds may be classified as viable in a germination test which provides optimum temperature, moisture and light conditions to the growing seedlings; however, they may not be capable of continuing growth and completing their life cycle under a wide range of field conditions. Generally, seeds start to lose vigor before they lose their ability to germinate; therefore vigor testing is an important practice in seed production programs. Vigor loss is due mainly to seed deterioration and aging, which starts as soon as the seed becomes physiologically mature. Its imperative, therefore, that seed be handled carefully to prevent accelerated reduction in performance through physical damage to cell membranes. This is particularly true of large seeds, such as pulses and legumes, where seed damage is the primary cause of deterioration. Using the appropriate vigor test for the crop kind under analysis will reveal the vigor of the seed. A vigor test is usually conducted in conjunction with a germination test because the latter is required before seed can be sold. Most producers want to know how their seed will perform at both ends of the spectrum, and these two tests will give them that information. Seed vigor has been determined by both germination rate and seedling growth rate. Germination rate measures the speed of germination which is commonly represented by time to 50% germination, while seedling growth rate is evaluated on a real time basis by measuring rate of elongation of the radical per unit time.

International Journal of Emerging Technology and Advanced Engineering


Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012) 2) Methods of measuring seed vigor: Vigor tests in general measure: i. Germination percentage of normal seedlings after imposition of stress such as cold or accelerated aging, ii. Biochemical tests such as tetrazolium or electrolyte conductivity and iii. Seedling growth. A vigor test cannot replace a germination test but rather supplements it with more information about seed quality. Naturally, the germination test will almost always have the higher result because the evaluation parameters are more forgiving than those used in the vigor test; so the germination test tells producers how their seed will perform in optimum conditions. A number of vigor tests proposed includes those listed in the following table II. RELATED WORK 1) Seedling and seed vigor: Substantial work for seed vigor and germination test using image processing has been reported. M. S. Howarth et.al. [1] Developed a machine vision system to measure root growth rate over the entire germination period. The objectives were to develop feature extraction techniques needed to measure seedling growth rate; to develop a machine vision system capable of monitoring seedling growth rate that included illumination, image acquisition, image processing, and pattern recognition requirements; and determine the reliability of the machine vision system as compared to human inspection performance. This technique provided an accurate representation of the growth rate as well as percent germination. For the root length measurement system they collected three runs of images for both lettuce and sorghum seeds followed by the root length computation for each seed by the machine vision system and the seed analysts. They compared the machine vision system with the seed analyst results. During testing, the seed analysts experienced several measurement errors. They had difficulty measuring curved roots and experienced errors in recording data. The average error between the system and manual measurement found was -0.13 cm for the lettuce test and -0.07 cm for the sorghum test. Percent germination computed by this system was found as 86.7% for the lettuce test and 96.7% for the sorghum test. The error reported between the manual and machine vision system measurements was small enough so that the machine vision system could be used to estimate seed vigor. The average time required to measure 10 roots per image was approximately 20 s. Future plans included enhancing the system so that more than 10 seeds can be analyzed during a test. Y. Sako1, et.al. [2] Presented a system for automated seed vigor assessment with a goal of cost and ease operation. The digital images of germinating seedlings captured by the system were processed by the computer to generate numerical values that collectively represent the quality of a seed lot (vigor index) based on sample mean of various statistics acquired from morphological features of the imaged seedlings. The system was tested on lettuce seedlings grown for three days in the dark. Accuracy of the system was tested by comparing the vigor index against a vigor index computed manually using individual seedling measurements. In the best case, the percentage difference between manual and computer determinations of the vigor index was found only 0.99 % for lot 1. 155

TABLE I
VIGOR TESTS

PHYSICAL SEED SIZE PHYSICAL


SOUNDNESS

PERFORMANCE FIRST COUNT SPEED


OF GERMINATION

STRESS COLD TEST COOL


GERMINATION TEST

BIOCHEMICAL GLUTAMIC
ACID DECARBOXYLA SE ACTIVITY TEST

COEFFICIENT
OF GERMINATION

BRICK GRAVEL
TEST

TETRAZOLIUM
TEST

SEEDLING
GROWTH RATE

PAPER
PIERCING TEST

RESPIRATION AND RQ MITOCHONDRI


AL ACTIVITY

SEEDLING DRY
WEIGHT

COMPACTED
SOIL

WET OR DRY
SOIL

ATP LEVEL MEMBRANE

PATHOGEN
INFESTED SOIL

INTEGRITY

ACCELERATED
AGING TEST

LOW OR HIGH P

International Journal of Emerging Technology and Advanced Engineering


Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012) In the worst case, the percentage difference was 14.71 % for lot 9. The imaging system captured images of seedlings from the side, enabling simultaneous measurements of both hypocotyls and radicles that provided accurate and reproducible seedling measurements for vigor assessment. R.Urena et.al.[3]presented an automatic system for monitoring seed germination by means of a software tool incorporating an artificial vision system to automate the data gathering process and a fuzzy logic-based classifier for automatic evaluation of germination quality. The system employed a color CCD camera to image trays in which seeds were planted in cells allowing the examination of the cells to search for germinated seeds. Then, automatic calculation of the leaf area of cotyledons and leaves was performed by using image processing algorithms (segmentation to obtain separate seedlings and description using leaf area). Finally the degree of development of the seedlings was classified using fuzzy logic and expert knowledge contained within a database, including tray number, sowing date, class, variety, elapsed days and GP given by experts. The fuzzy logic-based classifier used two indexes as inputs the GE and the DGR related to the uniformity and rate of the germination process to estimate the degree of development of the seedlings. By these means, the number of germinated seedlings and labels or indexes determining the quality of the germination process can be obtained by automatic measurement of the leaf area of the seedlings, using artificial vision, followed by classification of these measurements within a fuzzy logicbased framework. A.L. Hoffmaster et.al. [4] Developed a fast, objective, and reproducible image processing computer application to automatically assess the vigor of three-day-old soybean (Glycine max [L.] Merrill.) seedlings. The software extracted the digital images of soybean seed lots placed on a paper towel and converted into various digital representations. They used these representations to analyze the seedlings and segmented them into normal and abnormal categories. They further processed the normal seedlings and produced a one-pixel-wide summary structure of the shape of the seedling. From the summary structure, their software classified the seedlings into six type categories based on their shape (I, T, Y, P, A and Mtype seedlings). They processed each normal seedling to remove the cotyledon portion of the summary structure based on the type category it fell into. The length of each seedling in pixels was then computed by making use of the remaining summary structure, with the cotyledon removed. Then speed of the growth and uniformity of growth values were computed from these length measurements. 156 They normalized and combined these two values into a zero to 1,000 vigor index for the seed lot. The speed of growth value was dependent on the ratio of actual measured length to the maximum possible length for the seedlings found in the image. The uniformity of growth value was inversely proportional to the standard deviation of the length measurements for the seedlings. K. OAKLEY, et.al. [5] Used computer-aided image analysis to evaluate several measures of seedling growth in seed lots of Impatiens with different vigor, but high germination percentages. They selected seed lots from a single Impatiens cultivar based on their vigor using their Ball Vigor Index and verified using the saturated salts accelerated aging vigor test. They captured digital images from seeds germinated in Petri dishes placed on a flat-bed scanner and measured the seedling growth daily for four days following initial radicle protrusion using commercially available root length calculating software. Seedling growth (length and area), growth rate and a vigor index, were the measures that were evaluated. DELLAQUILA [6] proposed an overview of the experimental integration between the standard germination test and a computer-aided image analysis system with the aim to investigate the potential of the new technique in monitoring seed imbibition phases and radical elongation in broccoli, radish, lentil, lettuce and carrot seeds. The imaging project covered three major objectives: i) the development of a computer aided image analysis system to monitor seed imbibition, ii) the integration of germination test with seed image processing under a wide range of environmental conditions, such as NaCl stress (applied to broccoli), different temperature regimes (applied to broccoli and radish) or followed by controlled deterioration (applied to broccoli), iii) the definition of image analysis parameters in assessing early radicle elongation. The following germination parameters were estimated: germination percentage and the time in hours at which 10 % germination (T10) was reached, determined by extrapolating the corresponding germination curves. All the image analysis parameters such as Area, perimeter, length and width, (i.e. size change parameters), and the roundness factor were determined for each seed. Zhang Huaxian [7] focused on the research of rice (regular and hybrid) seed vigor determination using digital image processing. The TTC2, 3, 5-Triphenyl Tetrazolium Chloridedyeing method used in the seed vigor test have the advantages such as credible principles, exact results, simple method, timesaving and cheap cost.

International Journal of Emerging Technology and Advanced Engineering


Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012) Despite of the high veracity, manual determination of TTC dyeing results need high qualified technicians, and the technicians are prone to be tired. Combine with the machine vision systems, using digital image processing technology to deal with the results determination would settle this problem. In this paper, the author adopted both the TTC test to make the rice seed embryo dyed and image processing method. Thus making the seed vigor detection save more time, get more precise results, and also it declined to be applied on online and large quantity detection. The principle of TTC test is that live seed embryo has redox response in its respiration and TTC can be deoxidized to TTF (a red substance). Thus through the observation of the color and dyed area the vigor can be known indirectly. The author defined the vigor index and living ability index as follows:
Vigorindex theentirelydyedones / thewholeseeds 100 Livingabilityindex theentirelydyedones Mainstructuresdyed / thewholeseeds 100 ;

Using the TTC dyeing Method one can obtain a precise result of seed vigor, and it is also a very quick method compared to the standard germination test. It needs only 624 hours. Also, its results are precise. Many seed scientists in the world have compared the TTC test with the germination test, they find the result difference is no more than 1-2%. And through rice seed measurement, the result difference is in the range of 3 to 5%. N. Indrakumar Singh*et.al. [8] Presented the effects of seed size on seed quality within seed lot of pea crop. One commercial seed lot of pea cultivar (Pisum sativum) was size graded into three categories: large (seeds retained on a 2.36 mm screen), medium (seeds retained on a 2.0 mm screen) and small (ungraded). For statistical analysis, a completely randomized split plot design with twelve replicates (four replicates for each size grade) was used. Mean 1000 seed weight for each category was 250 g, 164 g and 126 g respectively. The graded seeds were tested for standard germination, vigor and field emergence. It was concluded from this experiment that positive relation was found between large seed grade and field emergence in all the test parameters. Large and medium size seeds have high seedling survival, growth and establishment under unfavorable condition, i.e. under field condition than small seeds. From the result of the AA test it can be suggested that to use as a carryover seed it would be suitable to use the medium (average) size seeds. These findings support the hypothesis of larger seeds those have superior performance to small seeds as the relation was found significantly positive. 157

It also concluded that a seed lot cannot be expressed as vigor by testing a sample of a lot as the lot consists of different sizes of seeds. McCormac et. al. (1990) investigated the use of an image analysis system for measuring the root length of lettuce using a slant board test. Root length was calculated by measuring the vertical distance from the zero axis which was manually located to the root tip. The root tip was detected by the imaging analyzer by single level thresholding. It was determined that the machine vision based system performed equally as well as the manual method. However, several problems were found to exist with this measurement technique. First, root length was only measured after the slant board test was completed. Secondly, only linear length was measured; therefore, the true length of the seedling root may not have been measured. Third, the length was measured from a stationary position. This position was the same for all seedlings. It is possible that this starting position was not the point separating root and hypocotyl. Since the method of human measurement was identical, the errors produced by these measurement characteristics were not realized. This source of error could potentially misrepresent vigor information. 2) Quality of seeds: Substantial work for identifying and classifying different grain varieties has been reported. Zayas et al. [9] illustrate the use of image analysis to discriminate between wheat and non-wheat components in a grain sample. They presented two methods, multivariate discriminate and a structural prototype method for pattern recognition. J. Paliwal et al. [10] also did a similar work for both bulk and single kernel seeds images by using flatbed scanner. N.S. Visen et al. [11] did a research to develop and optimize a technique for discrimination of various types of grains by extracting the morphological, texture, and color based features using images of single kernels and compare the classification accuracies using back propagation and specialist probabilistic neural network classifiers. N.S. Visen1 et al.[12] projected an image analysis system of bulk grain samples using neural networks. Identifying rice seeds varieties using neural network and image processing was projected by [13] LIU zhao-yan et al. They used a digital image analysis algorithm based on color and morphological features of rice seeds. M.A. Shahin et.al. [14] Tried to automate manual sieving procedure by using image analysis methods. They also explained about impossible situations used in most machine vision systems, where well-separated seeds used to image capturing under controlled laboratory conditions.

International Journal of Emerging Technology and Advanced Engineering


Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012) Therefore, they proposed seed size determination based on non-singulated samples. Jamuna KS et.al.[15] s work employed machine learning approach to classify the quality of seeds based on the different growth stages of the cotton crop. Machine learning is concerned with creating and using mathematical data structures that allow a computer to exhibit behaviour that would normally require a human. The following Machine learning techniques Nave Bayes Classifier, Decision Tree Classifier and Multilayer Perceptron were applied for training the model. Features were extracted from a set of 900 records of different categories to facilitate training and implementation. The performance of the model was evaluated using 10 -fold cross validation. The results obtained showed that Decision Tree Classifier and Multilayer Perceptron provide the same accuracy in classifying the seed cotton yield. The time taken to build the model is higher in Multilayer Perceptron as compared to the Decision Tree Classifier. This work can be further extended by incorporating other parameters such as varieties, spacing, water management, pest management. Classification of seed cotton can be carried out by employing support vector machines in future. L.A.I.Pabamalie et.al.[16] in their research focused on providing a better approach for identification of rice quality by using neural network and image processing concepts. In this research, a back propagation neural network with two hidden layers was developed for the quality classification. Thirty one texture and color features extracted from rice images were used for discriminate analysis. Tests on the system for the training and test sets show accuracy in between 94% to 68% for the four grades. S Borah et.al. [17] Examined about color detection in tea fermentation using an artificial neural network. They have attempted for testing the color with H (Hue), S (Saturation) and I (Intensity) color model using digital camera. P. M. Granitto et al. [18] studied the discriminating power of morphological, color and textural characteristics of weed seeds, which can be measured from video images. H. Rautio et.al. [19] Carried out another experiment for determining the average grain size by using the mathematical morphology and texture analysis. III. CONCLUSION In this paper classification is determined according to the use of image processing in seed technology and the type of result. An overview of the most representative methods is presented. 158 Commercial seed vigor assessment systems also exist to determine seed vigor such as Ball Vision Index (BVI) and paradigm system (PS), but these systems only examine seedling parts such as cotyledon area (BVI) or root length (PS).

REFERENCES
[1] M. S. Howarth, P. C. Stanwood MEASUREMENT OF SEEDLING GROWTH RATE BY MACHINE VISION, may-june-1993 [2] Y. Sako, M. B. McDonald, K. Fujimura, A. F. Evans, and M. A. Bennett A System for Automated Seed Vigor Assessment, 2001 [3] R. Urena, F. Rodrguez *, M. Berenguel A machine vision system for seeds germination quality evaluation using fuzzy logic, Computers and Electronics in Agriculture 32 (2001) 120 [4] A.L. Hoffmaster1*, K. Fujimura1, M.B. McDonald2, M.A. Bennett2 An Automated System for Vigor Testing Three-Day-Old Soybean Seedlings2003 [5] K. OAKLEY, S.T. KESTER AND R.L. GENEVE* Computer-aided digital image analysis of seedling size and growth rate for assessing seed vigor in Impatiens Oakley, K., Kester, S.T. and Geneve, R.L. (2004), Seed Sci. & Technol., 32, 907-915 [6] DELLAQUILA Application of a ComputerAided Image Analysis System to Evaluate Seed Germination under Different Environmental Conditions Ital. J. Agron., 8, 1, 51-62,2004 [7] Zhang Huaxian, Graduate Student, Cheng Fang, Professor, Determination of Rice Seed Vigor Using Digital Image Processing Technology 2007 ASABE Annual International Meeting Sponsored by ASABE Minneapolis Convention Center Minneapolis, Minnesota ,17 - 20 June 2007 [8] N. Indrakumar Singh*, Seema Ali* and J.S. Chauhan** Effect of Seed Size on Quality within Seed Lot of Pea and Correlation of Standard Germination, Vigor with Field Emergence Test Nature and Science, 2009;7(4), ISSN 15450740, [9] I. Zayas, Y. Pomeranz, and F. S. Lai, Discriminate between wheat and non-wheat components in a grain Sample, Cereal Chem., vol. 66, no.3, 1989. [10] J. Paliwal,M. S. Borhan and D. S. Jayas, Classification of cereal grains using a flatbed scanner, Can Biosyst Eng,vol. 46, 2004. [11] N. S. Visen, D. S. Jayas, J. Paliwal, and N. D. G.White, Comparison of two neural network architectures for Classification of singulated cereal grains, Can. BioSyst. Eng, vol. 46, 2004. [12] N. S. Visen1, J. Paliwal1, D. S. Jayas1, and N. D. G.White2, Image analysis of bulk grain samples using Neural networks, Can. BioSyst. Eng, vol. 46, 2004.

International Journal of Emerging Technology and Advanced Engineering


Website: www.ijetae.com (ISSN 2250-2459, Volume 2, Issue 4, April 2012)
[13] L. Zhao yan, C. Fang, Y. Yibin, and R. Xiuqin, Identification of rice seed varieties using neural network, Journal of Zhejiang University SCIENCE, ISSN 1009-3095, September 2005. [14] M. A. Shahin and S. J. Symons, Seed sizing from images of nonsingulated grain samples, Can. BioSyst. Eng, vol. 47, 2005. [15]Jamuna KS, Karpagavalli S, Vijaya MS ,Revathi P, Gokilavani S, Madhiya EClassification of Seed Cotton Yield based on the Growth stages of Cotton crop using Machine Learning Techniques 2010 International Conference on Advances in Computer Engineering [16]L.A.I.Pabamalie, H.L.Premaratne A Grain Quality Classification System,2010 [17] S. Borah, M. Bhuyan and H. Saikia, ANN based Color Detection in Tea Fermentation. [18] P. M. Granitto,H. D. Navone, P. F. Verdes, and H. A. Ceccatto, Automatic identification of weed seeds by color image processing. [19] H. Rautio and O. Silvn, Average Grain Size Determination using Mathematical Morphology and Texture Analysis.

159

You might also like