Image based oil palm fruit bunch growth modeling for harvesting operation

Oil Palm Fresh Fruit Bunch (FFB) need to be harvested at the optimum maturity stage to optimize the quality of palm oil. Currently the oil palm harvester determines the FFB maturity based on natural indicators such as FFB color appearance and number of FFB loose fruit drops under the tree. During ex...

Full description

Saved in:

Bibliographic Details
Main Author:	Mohd Kassim, Muhamad Saufi
Format:	Thesis
Language:	English
Published:	2013
Subjects:	Oil palm Palm oil industry - Waste disposal
Online Access:	http://psasir.upm.edu.my/id/eprint/67688/1/ITMA%202013%207%20IR.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!

id	my-upm-ir.67688
record_format	uketd_dc
institution	Universiti Putra Malaysia
collection	PSAS Institutional Repository
language	English
topic	Oil palm Palm oil industry - Waste disposal
spellingShingle	Oil palm Palm oil industry - Waste disposal Mohd Kassim, Muhamad Saufi Image based oil palm fruit bunch growth modeling for harvesting operation
description	Oil Palm Fresh Fruit Bunch (FFB) need to be harvested at the optimum maturity stage to optimize the quality of palm oil. Currently the oil palm harvester determines the FFB maturity based on natural indicators such as FFB color appearance and number of FFB loose fruit drops under the tree. During executing the harvesting operation the harvester need to search for a ripe FFB and at the same time carrying the harvesting pole. Tedious harvesting operation degrades the consistency of their judgment. The harvested FFB must be graded at the oil palm mill to separate into groups of maturity level according to the standard. In this research, the development of FFB from the anthesis to harvesting stage was monitored by using a handy digital camera over a period of eight months. A computer application called Growth Table was developed to manage the FFB digital images and ease the process of grouping the digital images into 25 groups of FFB maturity stages. The digital images of the FFB were processed by using digital image processing techniques to extract the color information that represent the maturity stages. Two types of color spaces were investigated, HSV(Hue, Saturation and Value) and RGB (Red, Green and Blue) color space. In HSV color space only Hue component was considered to extract maturity information. During the process, a clustering technique was used to separate every single FFB image into three color clusters that represent three FFB features which were Fruitlet, Brown Spine and Green spine. As a result from the analysed image information and tabulated data in Growth Table, a relationship of FFB features color changing and maturity stages were investigated. The Growth Model of the above relationship was developed. During the process it was found that the FFB grow in stages. In Hue color component, the FFB was found to grow in three major stages. First Major Growth Stages(hue) (FMGSh) was from week 0 to 5, Second Major Growth Stages(hue) (SGMSh) was from week 5 to 14 and Third Major Growth Stages(hue) (TMGSh) from week 15 to 24. FFB development in RGB color space was found to have two major growth stages. First Major Growth Stages (FMGS) from week 0 to 5 and Second Major Growth Stages (SMGS) were from week 6 to 24. From the regression analysis, linear models and multiple linear models of each major growth stages was determined to develop the Growth Models. Predicted maturity stages data using the developed Growth Models were validated with the actual maturity stage as determined by using the Growth Table. In term of the accuracies of predicted data as compared with the actual data, the best Hue model had an R2= 0.95 for the third growth stage while the best RGB model had an R2= 0.9 for first growth stage. The processed information by using the developed Growth Model also enables the development of FFB Harvesting Model. The data from Harvesting Model can be used to generate a graphical oil palm leaf spiral that mapped the location of FFB in relation with the location of oil palm leaves. FFB production can be monitored by observing the presence of FFB at oil palm 17th leaves position that is the beginning of anthesis phase to oil palm leaves at 32nd where FFB is at optimum maturity stage. A GIS map displaying the location of matured FFB on the tree and the maturity stages of the FFB can be generated. The GIS map can be used as a support system in site specific harvesting operation. This Harvesting Model enables site specific harvesting at optimum maturity stage, overcome losses due to uncollected loose fruits. Growth Model has a potential to eliminate FFB screening process at oil palm mill level. Harvesting Model can be a tool of choice for better harvesting scheduling and can be a good tool to predict FFB yield. The developed models from this research have a high potential to improve oil palm field management as well as oil palm mill.
format	Thesis
qualification_level	Doctorate
author	Mohd Kassim, Muhamad Saufi
author_facet	Mohd Kassim, Muhamad Saufi
author_sort	Mohd Kassim, Muhamad Saufi
title	Image based oil palm fruit bunch growth modeling for harvesting operation
title_short	Image based oil palm fruit bunch growth modeling for harvesting operation
title_full	Image based oil palm fruit bunch growth modeling for harvesting operation
title_fullStr	Image based oil palm fruit bunch growth modeling for harvesting operation
title_full_unstemmed	Image based oil palm fruit bunch growth modeling for harvesting operation
title_sort	image based oil palm fruit bunch growth modeling for harvesting operation
granting_institution	Universiti Putra Malaysia
publishDate	2013
url	http://psasir.upm.edu.my/id/eprint/67688/1/ITMA%202013%207%20IR.pdf
_version_	1747812500720582656
spelling	my-upm-ir.676882019-03-22T07:44:08Z Image based oil palm fruit bunch growth modeling for harvesting operation 2013-06 Mohd Kassim, Muhamad Saufi Oil Palm Fresh Fruit Bunch (FFB) need to be harvested at the optimum maturity stage to optimize the quality of palm oil. Currently the oil palm harvester determines the FFB maturity based on natural indicators such as FFB color appearance and number of FFB loose fruit drops under the tree. During executing the harvesting operation the harvester need to search for a ripe FFB and at the same time carrying the harvesting pole. Tedious harvesting operation degrades the consistency of their judgment. The harvested FFB must be graded at the oil palm mill to separate into groups of maturity level according to the standard. In this research, the development of FFB from the anthesis to harvesting stage was monitored by using a handy digital camera over a period of eight months. A computer application called Growth Table was developed to manage the FFB digital images and ease the process of grouping the digital images into 25 groups of FFB maturity stages. The digital images of the FFB were processed by using digital image processing techniques to extract the color information that represent the maturity stages. Two types of color spaces were investigated, HSV(Hue, Saturation and Value) and RGB (Red, Green and Blue) color space. In HSV color space only Hue component was considered to extract maturity information. During the process, a clustering technique was used to separate every single FFB image into three color clusters that represent three FFB features which were Fruitlet, Brown Spine and Green spine. As a result from the analysed image information and tabulated data in Growth Table, a relationship of FFB features color changing and maturity stages were investigated. The Growth Model of the above relationship was developed. During the process it was found that the FFB grow in stages. In Hue color component, the FFB was found to grow in three major stages. First Major Growth Stages(hue) (FMGSh) was from week 0 to 5, Second Major Growth Stages(hue) (SGMSh) was from week 5 to 14 and Third Major Growth Stages(hue) (TMGSh) from week 15 to 24. FFB development in RGB color space was found to have two major growth stages. First Major Growth Stages (FMGS) from week 0 to 5 and Second Major Growth Stages (SMGS) were from week 6 to 24. From the regression analysis, linear models and multiple linear models of each major growth stages was determined to develop the Growth Models. Predicted maturity stages data using the developed Growth Models were validated with the actual maturity stage as determined by using the Growth Table. In term of the accuracies of predicted data as compared with the actual data, the best Hue model had an R2= 0.95 for the third growth stage while the best RGB model had an R2= 0.9 for first growth stage. The processed information by using the developed Growth Model also enables the development of FFB Harvesting Model. The data from Harvesting Model can be used to generate a graphical oil palm leaf spiral that mapped the location of FFB in relation with the location of oil palm leaves. FFB production can be monitored by observing the presence of FFB at oil palm 17th leaves position that is the beginning of anthesis phase to oil palm leaves at 32nd where FFB is at optimum maturity stage. A GIS map displaying the location of matured FFB on the tree and the maturity stages of the FFB can be generated. The GIS map can be used as a support system in site specific harvesting operation. This Harvesting Model enables site specific harvesting at optimum maturity stage, overcome losses due to uncollected loose fruits. Growth Model has a potential to eliminate FFB screening process at oil palm mill level. Harvesting Model can be a tool of choice for better harvesting scheduling and can be a good tool to predict FFB yield. The developed models from this research have a high potential to improve oil palm field management as well as oil palm mill. Oil palm Palm oil industry - Waste disposal 2013-06 Thesis http://psasir.upm.edu.my/id/eprint/67688/ http://psasir.upm.edu.my/id/eprint/67688/1/ITMA%202013%207%20IR.pdf text en public doctoral Universiti Putra Malaysia Oil palm Palm oil industry - Waste disposal

Image based oil palm fruit bunch growth modeling for harvesting operation

Similar Items