Şeker pancarı tarımında yabancı ot mücadelesi için değişken düzeyli herbisit uygulama parametrelerinin yapay sinir ağlarıyla belirlenmesi
Yükleniyor...
Dosyalar
Tarih
2013-03-06
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Selçuk Üniversitesi Fen Bilimleri Enstitüsü
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Bu çalışmada, şeker pancarı üretimi yapılan tarlada, sıra arası ve sıra üzerinde bulunan yabancı otları tespit ederek ilaçlanması için üzerinde ilaçlama ünitesi, kontrol ünitesi, bilgisayar ve kameraların bulunduğu bir ilaçlama robotu tasarlanarak imal edilmiştir. Geliştirilen bu robot ile sıra arasındaki yabancı otlar renk bilgisine, sıra üzerindeki yabancı otlar ise şekil ve renk bilgisine göre tespit edilerek üzerilerine ilaçlama sıvısı (mürekkepli su) uygulanmıştır. Sıra arasındaki yabancı otlar görüntü işleme teknikleri ile sıra üzerindeki yabancı otlar ise görüntü işleme ve yapay sinir ağları kullanılarak tespit edilmiştir. Şeker pancarı (Beta vulgaris) ve şeker pancarı tarlasında sık rastlanan yabancı otlardan sirken (Chenopodium album), yabani marul (Lactuca serriola), deve dikeni (Carduus nutans) ve darıcan (Echinochloa crus-galli) otları tarlada fotoğraflanmıştır. Bu fotoğrafların çıktısı alındıktan sonra PVC ile kaplanarak denemelerde kullanılan kültür bitkisi ve yabancı ot resimleri elde edilmiştir. Laboratuar koşullarında yürütülen denemelerde ilaçlama robotunun 8 farklı hız değerinde yabancı otlar üzerine uygulanan ilaçlama sıvısının yabancı otu örtme yüzdeleri, uygulanan ilaçlama sıvısı miktarları ve sistemin yabancı otu tanıma performansı deneyleri yapılmıştır. İlaçlama robotunun hızının artırılması, yabancı otlar üzerine uygulanan ilaçlama sıvısı miktarında bir azalmaya neden olduğu gibi, uygulanan ilaçlama sıvısının bitki yüzeyini örtme yüzdelerini de azalttığı görülmüştür. İlaçlama robotu üzerindeki meme yüksekliği azaldıkça uygulanan ilaçlama sıvısının bitkiyi örtme yüzde değerinin ve uygulanan ilaçlama sıvısı miktarının azaldığı tespit edilmiştir. Bu çalışmada hassas ilaçlama robotu ile yapılan uygulamada, uygulanan ilaçlama sıvısı miktarında geleneksel ilaçlama uygulamasına göre %53.96 değerinde tasarruf edilmiştir. Böylece, sistem yabancı otları tespit ederek tarlanın geneli yerine sadece yabancı otlara ilaçlama yapacağından insan, hayvan ve çevre sağlığının korunması yanında ilaçlama maliyetlerinin düşürülmesi sağlanabilinecektir.
In this study, a disinfecting robot having a disinfection unit, control unit, a computer and cameras was designed and made with the purpose of detecting and disinfecting weeds existent in interrows and rows in a sugar beet field. Through this robot, weeds in interrows were detected according to their colour and weeds on rows were detected according to both their shape and colour, and then disinfecting liquid (diluted ink) was applied on them. Weeds in interrows were detected through image processing techniques and weeds on rows were detected by the help of image processing and artificial neural networks. Sugar beet (Beta vulgaris) and such frequently found weeds in sugar beet fields as lamb's quarters (Chenopodium album), prickly lettuce (Lactuca serriola), musk thistle (Carduus nutans), and cockspur grass (echinochloa crus-galli) were photographed in the field. These photographs were printed out and covered with plastics so that photos of sugar beet and weeds used in tests were obtained. Tests were carried out in laboratory conditions. During tests, at eight different speed values of the disinfecting robot, the percentage of disinfecting liquid?s covering weed, the amounts of disinfecting liquid applied on weeds, the weed recognition performance of the system were experimented. While an increase in the speed of the disinfecting robot caused a decrease in the amount of disinfecting liquid applied on weeds, it was also observed that disinfecting liquid decreased the percentage of covering plant surface. It was detected that as the distance between the nozzle on disinfecting robot and the ground was shortened, the percent of disinfecting liquid?s covering the plant was lowered and the amount of disinfecting liquid applied on weeds decreased. In this study, in an implementation carried out with a sensitive disinfecting robot, when compared to conventional pulverizing methods there was a saving in the amount of the applied disinfecting liquid by %53.96. In this way, since the system is to detect weeds and implement disinfection not throughout the whole field but only on the weeds, human, animal and environmental health will be able to be preserved and disinfection costs will be reduced.
In this study, a disinfecting robot having a disinfection unit, control unit, a computer and cameras was designed and made with the purpose of detecting and disinfecting weeds existent in interrows and rows in a sugar beet field. Through this robot, weeds in interrows were detected according to their colour and weeds on rows were detected according to both their shape and colour, and then disinfecting liquid (diluted ink) was applied on them. Weeds in interrows were detected through image processing techniques and weeds on rows were detected by the help of image processing and artificial neural networks. Sugar beet (Beta vulgaris) and such frequently found weeds in sugar beet fields as lamb's quarters (Chenopodium album), prickly lettuce (Lactuca serriola), musk thistle (Carduus nutans), and cockspur grass (echinochloa crus-galli) were photographed in the field. These photographs were printed out and covered with plastics so that photos of sugar beet and weeds used in tests were obtained. Tests were carried out in laboratory conditions. During tests, at eight different speed values of the disinfecting robot, the percentage of disinfecting liquid?s covering weed, the amounts of disinfecting liquid applied on weeds, the weed recognition performance of the system were experimented. While an increase in the speed of the disinfecting robot caused a decrease in the amount of disinfecting liquid applied on weeds, it was also observed that disinfecting liquid decreased the percentage of covering plant surface. It was detected that as the distance between the nozzle on disinfecting robot and the ground was shortened, the percent of disinfecting liquid?s covering the plant was lowered and the amount of disinfecting liquid applied on weeds decreased. In this study, in an implementation carried out with a sensitive disinfecting robot, when compared to conventional pulverizing methods there was a saving in the amount of the applied disinfecting liquid by %53.96. In this way, since the system is to detect weeds and implement disinfection not throughout the whole field but only on the weeds, human, animal and environmental health will be able to be preserved and disinfection costs will be reduced.
Açıklama
Anahtar Kelimeler
Hassas tarım, Görüntü işleme, Yapay sinir ağları, Şeker pancarı tarımı, Yabancı ot mücadelesi, Precision agriculture, Image processing, Artificial neural networks, Sugar beet farming, Weed control
Kaynak
WoS Q Değeri
Scopus Q Değeri
Cilt
Sayı
Künye
Sabancı, K. (2013). Şeker pancarı tarımında yabancı ot mücadelesi için değişken düzeyli herbisit uygulama parametrelerinin yapay sinir ağlarıyla belirlenmesi. Selçuk Üniversitesi, Yayımlanmış yüksek lisans tezi, Konya.
