Abstract
Today agriculture field’s demands to develop such an intelligent system those provide accurate and timely information for an estimation of crop productivity. This paper designed an automated decision support system to estimate sunflower crop productivity information with interface between camera and computer software. The earlier steps of system generate overlapped flower yield information and latter steps count the seed from the flower head. Some beautiful flowers in the nature have Fibonacci relationship in their seeds pattern, i.e. sunflower, pineapple etc. The implementation parts based on two color model RGB and HSV. HSV provide better results for overlapped flower. The technique use image segmentation, morphological operation for overlapped flower count and edge detection for seed count.
Access provided by Autonomous University of Puebla. Download conference paper PDF
Similar content being viewed by others
Keywords
1 Introduction
Study of flowers is done in Floriculture [1]. The study of modern flower crops comes under the floriculture which is sub-branch of horticulture [1]. Yield estimation is the main part of the agriculture’s accuracy. The precision agriculture has main application of the yield estimation. Precision agriculture is the early estimation of the crop for preplanning and decision of every stage of crop. The entire problem in every stage of flowers can be solved by precision agriculture using computer & machine vision based decision with full automation. This paper presented the sunflower yield calculation of crop from flowers shape analysis [2]. The man made estimation system gives better results even if there is variation in fields or weather conditions [3,4,5,6,7]. For developing a new computing algorithm basic yield estimation model is inferred in Fig. 1 below.
Object detection, shape detection and texture detection are the three main parts of counting algorithm [8,9,10,11]. It is easy to estimate the object from binary image in matlab. Hence, firstly it is important to use segmentation technique for extracting flower form the background and convert in to binary image [12]. In the fifteenth century scientist Leonardo discovered Fibonacci series [13]. The sun magic flower known as sunflower. It has clockwise and anticlockwise spires. These spires have number relationship (i.e. 34 anticlockwise and 55 clockwise or 55 anticlockwise and 89 clockwise). For any size of sunflower they come with Fibonacci number, even if head size of sunflower very. It is very complex to count number of seeds from a sunflower because many seeds presented in both spires and they intersect. In Fig. 2 show the algorithm steps of sunflower yield estimation.
2 Development Algorithm
The Development steps of proposed algorithm are divided into two parts, first for flower count and second for seed count:
The flower count step includes 6 steps, in which input is taken as RGB image and output is generated in the form of flower count. In next seed count steps includes the cropping the sunflower head followed by seed count operation.
2.1 For Flower Count
-
Step 1: First of all Capture RGB images from the field of the sunflower
The first step of yield estimation of sunflower is capturing the image of sunflowers from the sunflower field. The sunflower image is captured by the high-quality camera which is stationary. The image captured by camera is the RGB image. The RGB image has the basic color of any object like Red, Green, and Blue. Future this RGB image to pass to next step which remove the noise from the RGB image.
-
Step 2: Remove the noise from the captured image
Salt and pepper noise is common in captured image by camera. So remove the Salt and pepper noise by the Median filter. The output pixels of the image are determined by calculating the median of the neighborhood pixels and replace the pixel by the calculated median pixel. In this filter, the median value is placed on image pixel value.
-
Step 3: Convert the RGB capture image to HSV image
Different flowers have the different color. The color extraction has a range of hues value because the flowers contain shades of color. After the color transformation, the hue range is calculated. The HSV color model used for flower extraction. The transformation of RGB image gives better segmentation. The segmentation of the flowers is better using HSV color model then compared to RGB color model.
-
Step 4: Generate the binary image
After detection of flowers, they are extracted from the background using Otsu thresholding technique segmentation process. After the segmented image is called the binary image in which flower region is white and the background is black and vice versa. It is easy to count objects from the binary image.
-
Step 5: Apply Morphological operation on the binary image
The image is reconstructed using the finite number of time operation i.e. Dilation, Erosion, Opening, and Closing after the morphological operation.
-
Step 6: Apply edge detection process followed by circular fitting and flower count
The basic process of detection detects the boundaries of different object and outline of an object and background of the image and indicates the overlapping object boundaries. The image segmentation using the canny edge. Find the number of circles and their radius by using the imfindcircles commands. The number of circles is equal to the number of the flower head and the size of each flower is calculated in inch from the radius.
2.2 For Seeds Count
-
Step 1: Head selection after all steps of Flower Count Steps then cropping
After the circular fitting, the circle marks on the head of the sunflower then count the sunflower head and find an estimation of the sunflower cropping. Heads of the sunflowers are not the same size. Different sunflower has different head size then select the different sunflower head for next step seed count.
-
Step 2: Seed Counting and estimation of the crop of sunflower
After the head count, then select the sunflower head of different size. After head selection, the seeds are count form different heads of sunflower and find the average weight of the seeds which give the crop estimation of sunflower.
3 Experimental Results
During the image capturing process there is some chance to add the noise due to any regions like environment condition etc. Hence here is need to remove such kind of noise [14,15,16]. RGB to HSV conversion is necessary because the flower area can be detected on the basis of hue value [17, 18]. It is easy to extract the different regions by using segmentation technique [19]. The circular fitting or curve fitting tool box used to count the number of flower [20,21,22]. It provides the facility to create, modify and access the fitting objects [20,21,22]. The methods command of MATLAB results the curve fitting objects.
For example,
f = fittype (‘a * x2 + b * exp (n * x)’); methods (f)
The number of objects or circles gives the result Flower yield calculation.
Figure 3 infers the results obtained after execution of flower count steps. From the results it is clear that accuracy of flower count is near about 92%.
4 Accuracy of Flower Count Steps
The number of flower count is equal to the result of the length function of MATLAB.
For example:
length (centers);
The accuracy of proposed flower count steps can be computed by comparing it with the manual count.
Here,
-
AFC = Algorithmic flower count
-
MC = Manual count
In Table 1 shows the Summary for the accuracy of different kinds of flowers.
From above table it can be concluded that the flower count steps of proposed algorithm provides the 89.896% or 90% accuracy. For sunflower it results accuracy approximately 92%. Input and output image from figure no 2 it is clear that head count is 22 out of 24, here the accuracy is 91.67%.
5 Head Size Based on Circular Fitting
Figure 2 shows overlapped flower count or flower head count. Sunflower is found in many verities but yellow sunflower is common. In agriculture the head size is normally measured in inch. It very from 4 inch to 12 inch. The maximum head size exists up to 27 inch. Table 2 shows the radius of sunflowers reflected as result, from the output image in Fig. 2 and its equivalent full head size in inch.
1 inch = 2.54 cm or 1 cm = (1/2.54 = 0.393701) inch
In this paper we categorize three kinds of sunflower heads; these are Large, Medium and Small [23].
So close picture of such kind of head is taken and count the seed of individual head using edge detection.
Apply the following Mathematical formula to seed estimation:
Here,
-
SE: Total seed count
-
FC: flower count
-
SHL: Number of seed in large head (greater than 10″).
-
SHM: Number of seed in medium head (7″ >= and <=10″).
-
SHS: Number of seed in small head (less than 7″).
Here we can categorize the head in more, according to size which give more accurate yield results.
Following Fig. 4 infer the edge detection.
Table 3 shows the Results of average weight yield. From the Pearson Education, “Data analysis on the Web” the average weight of a seed is 142 mg [24].
So yield estimated weight can be derived from it.
6 Accuracy of Crop Estimation
The accuracy of crop estimation of sunflower can be computed by comparing with the manual count average weight
Here,
-
ACW = Algorithmic Count Weight
-
MCW = Manual Count Weight
In the Table 4, the input image has 24 head and output image has 22 head, then the weight accuracy of 22 head out of 24 is 92.28%.
7 Conclusion
This paper presents the sunflower crop yield estimation technology with the help of the application of image processing. To comparing our results on the basis of seed weight [24]. Our results meet the accurate crop productivity. it can be concluded that the technique provides the accurate estimation of sunflower weighted crop. The proposed technique is used for precise agriculture. In future scope, we will focus on other types of crop.
References
Chadha, K.L.: Horticulture: new avenues for growth. The Hindu Survey of Indian Agriculture, 155–160 (1999)
Wang, Q., Nuske, S., Bergerman, M., Singh, S.: Automated crop yield estimation for apple orchards. In: 13th International Symposium on Experimental Robotics (ISER 2012), pp. 1–15 (2012)
Stajnko, D., Rakun, J., Blanke, M.: Modelling apple fruit yield using image analysis for fruit colour, shape and texture. Europ. J. Hortic. Sci. 74, 260–267 (2009)
Bairwa, N., Agrawal, N., Gupta, S.: Development of counting algorithm for overlapped agricultural products. In proceeding of National Conference on Recent Advances in Wireless Communication and Artificial Intelligence (RAWCAI-2014) (2014)
Zhou, R., Damerow, L., Sun, Y., Blanke, M.: Using colour features of cv. ‘Gala’ apple fruits in an orchard in image processing to predict yield. Precis. Agric. 13, 568–580 (2012)
Stajnko, D., Rozmana, Č., Pavloviča, M., Beber, M., Zadravec, P.: Modeling of ‘Gala’ apple fruits diameter for improving the accuracy of early yield prediction. Sci. Hortic. 160, 306–312 (2013)
Kelman, E., Linker, R.: Vision-based localisation of mature apples in trees images using convexity. Biosyst. Eng. 118, 174–185 (2014)
Bairwa, N., Agrawal, K.N.: Counting algorithm for flowers using image processing. Int. J. Eng. Res. Technol. (IJERT) 3, 775–779 (2014)
Gil, J., Kimmel, R.: Efficient dilation, erosion, opening and closing algorithms. In: Goutsias, V.J., Vincent, L., Bloomberg, D. (eds.) Mathematical Morphology and its Applications to Image and Signal Processing, Palo-Alto, USA, June 2000, pp. 301–310. Kluwer Academic Publishers (2000)
Haralick, R.M., Shanmugam, K., Dinstein, I.: Textural Features for image classification. IEEE Trans. Syst. Man Cybernatics 3(6), 610–621 (1973)
Guru, D.S., Sharath, Y.H., Manjunath, S.: Texture features and KNN in classification of flower images. IJCA (2010). Special Issue on Recent Trends in Image Processing and Pattern Recognition RTIPPR
Patel, H.N., Patel, A.D.: Automatic segmentation and yield measurement of fruit using shape analysis. Int. J. Comput. Appl. 45(7) (2012). (0975 – 8887)
Aggelopoulou, A.D., Bochtis, D., Fountas, S., Swain, K.C., Gemtos, T.A., Nanos, G.D.: Yield prediction in apple orchards based on image processing. Precis. Agric. 12, 448–456 (2011)
Sharma, V.K., Srivastava, D.K., Mathur, P.: Efficient image steganography using graph signal processing. IET Image Process. 12(6), 1065–1071 (2018)
Kaur, S., Kaur, S.: Performance evaluation of different filters in image denoising for different noise. Int. J. Innov. Res. Sci. Eng. Technol. 5(7) (2016). ISSN (Online) 2319-8753, ISSN (Print) 2347-6710
Singh, P., Shree, R.: A comparative study to noise models and image restoration techniques. Int. J. Comput. Appl. 149(1) (2016). (0975 – 8887)
Saravanan, G., Yamuna, G., Nandhini, S.: Real time implementation of RGB to HSV/HSI/HSL and its reverse color space models. In: International Conference on Communication and Signal Processing, India, 6–8 April 2016 (2016)
Nnolim, U.A.: Design and implementation of novel, fast, pipelined HSI2RGB and log-hybrid RGB2HSI color converter architectures for image enhancement. Microprocess. Microsyst. 39(4–5), 223–236 (2015)
Nilsback, M.E., Zisserman, A.: Delving into the whorl of flower segmentation. In the Proceedings of British Machine Vision Conference, vol. 1, pp. 27–30 (2004)
Islam, S., Ahmed, M.: A study on edge detection techniques for natural image segmentation. Int. J. Innov. Technol. Explor. Eng. (IJITEE) 2(3) (2013). ISSN 2278-3075
Wang, H.: Robust statistics for computer vision: model fitting, image segmentation and visual motion analysis, Ph.D thesis, Monash University, Australia (2004)
Yuen, H.K., Princen, J., Illingworth, J., Kittler, J.: A comparative study of hough transform methods for circle finding. Image Vis. Comput. 8, 71–77 (1990)
Ahmad, S.: Environmental effects on seed characteristics of sunflower. J. Agron. Crop Sci. 187(3), 213–216 (2001)
StatCrunch: Data Analysis on Web. College of Western Idaho, Pearson Education 2017–2018
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Rathore, H., Sharma, V.K., Chaturvedi, S., Sharma, K.D. (2019). Overlapped Sunflower Weighted Crop Yield Estimation Based on Edge Detection. In: Luhach, A., Singh, D., Hsiung, PA., Hawari, K., Lingras, P., Singh, P. (eds) Advanced Informatics for Computing Research. ICAICR 2018. Communications in Computer and Information Science, vol 955. Springer, Singapore. https://doi.org/10.1007/978-981-13-3140-4_2
Download citation
DOI: https://doi.org/10.1007/978-981-13-3140-4_2
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-13-3139-8
Online ISBN: 978-981-13-3140-4
eBook Packages: Computer ScienceComputer Science (R0)