棉 花 学 报 Cotton Science 2007，19(2)：106-113
Study on Cotton N Status Diagnosis Using Digital Image Processing
WANG Xiao-jing1，2，ZHANG Yan1*，LI Pan1，HOU Xiu-ling1，2，FENG Gu2
(1. Soil and Fertilizer Institute，Xinjiang Academy of Agricultural Sciences，Urumqi 830000，China；2. College of Resources and Environment，China Agricultural University，Beijing 100094，China)
Abstract：The common methods including testing of total N、stem tissue nitrate and SPAD readings were used to diagnose the nitrogen (N) status of cotton and recommend the nitrogen application rate. However ，due to labor requirements and high costs，these methods can hardly be applied to the cotton in field condition. In recent years，with rapid development of remote sensing nearing the ground，digital image analysis method combined with plant test and yield measurement has become another choice in crop nutrient diagnosis. Because N of plant shows obvious effects on chlorophyll constitutes and content directly，nutrition could be detected through digital image analysis of plant canopy. Large quantities of tests have been studied on spectrum character of plant canopy. But few of that were studied on cotton. Therefore，we are trying to research on spectrum character of cotton canopy through digital image analysis method for N status diagnosis.
In this study，digital image analysis method，which combined with soil and plant N rapid test，was studied for cotton N status diagnosis. The experiment was conducted within a multi-factorial field trial with three levels of N under channel irrigation at Fenshou farm，Awati county，Xinjiang. The possibility of using digital image analysis method for plant N status diagnosis was studied. The results show significant relationship between color index of plant canopy and N rate，yield，total N，plant nitrate concentration and biomass. Compared with other canopy color index，the normalized intensity of redness and blueness R/(R+G+B)，B/(R+G+B) was much better because of its significant lineal inverse relationships correlations with N rate，yield at boll stage. These could be suitable indicators for N status diagnosis. The plant canopy indexes of G and G/R were good correlated with total N content， nitrate concentration and biomass. They could be used for N status diagnosis at boll stage.
Results showed that digital image analysis method combined with soil and rapid test of petiole nitrate could be used for nitrogen fertilization rate in field condition. In this field experiment， when G/(R+G+B) was less than 1.16，cotton was in N stress state，did need more N topdressing at boll stage in a degree；when G/(R+G+B) was 1.16-1.18，cotton was in optimum N supply state；and when N fertilization rate before boll stage exceed what the plants actually need，G/(R+G+B) would be less than 1.155 and cotton was in surplus N at this time. However，many factors such as water stress，crop variety ，field weed，insect pests or any other factors affecting weather condition could influence leaf color of cotton canopy. These effects they brought were unavoidable in field condition. Consequently，simply depending on results from canopy image information is not enough for nutrient diagnosis and N fertilization recommendation. It is important that the integration of soil or rapid plant test and digital image analysis method should be emphasized. On all accounts，more inspection and test should be devoted to application of digital image analysis method on cotton nutrition diagnosis and nutrient resources management. [Full Text，3261KB]
|