Analysis Of Extraction Method Of Garlic Planting Area In Qixian County Based On ENVI

Garlic is a significant economic crop.It is an important edible seasoning and vegetable with high medicinal value.China is the world's leading garlic producer,consumer,and exporter,whose production occupies the world's total output accounting for 90%.2017,harvested area reaching 8 million Mu.And the production of Qixian amounted to 900,000 tons with 700,000 Mu,being one of the major garlic producing areas in China.In order to strengthen the production and management,adjust the agricultural structure,and formulate a scientific and rational agricultural policy,we should know about the planting area timely.In the past,the statistics of crop planting areas mainly depended on artificial field surveys to complete the statistics by layer and layer.This kind of statistical method was time-consuming and laborious with amounts of statistical errors.Using remote sensing technology,large-area surface information can be obtained in different spatial and temporal resolutions.The extraction of target information can eliminate human factors from interference in the condition of scientificity and accuracy.There are little studies on remote sensing extraction of garlic planting area in the domestic and overseas which results in the lacking of technology for planting.The main area of garlic production area is in a high degree of fragmentation,the planting area is more dispersed and the intercrop intercropping is serious.It is difficult to extract and figure out the garlic planting area.Based on the ENVI remote sensing image processing platform,this paper selects the image data of high score No.1 and Landsat-8 as the data source,and makes a comparative study on the extraction method of garlic area in the study area.This paper introduces the general situation of the research area,the acquisition of remote sensing data,the preprocessing of remote sensing data,the method of supervision and classification,and the evaluation criteria of classification accuracy,and compares the recognition accuracy of different data sources,different supervised classifications and different recognition schemes,and obtains the best method for the extraction of garlic area by remote sensing.The results showed that:1)The overall classification accuracy of Landsat-8 was higher than that of the high score one,but the high score one had higher accuracy in identifying garlic.2)Single phase image interpretation,January is the best time to extract garlic area.The accuracy of garlic area interpretation in multi time NDVI time series is not very different from that in single phase.Considering the calculation and calculation efficiency,it is concluded that single time phase data is the best method for garlic area interpretation.3)The extraction of garlic area in Qixian County based on single temporal images is the best method.The extraction of garlic area based on multi temporal NDVI sequence is the best method of support vector machine and maximum likelihood method.In the comprehensive evaluation,support vector machine is the highest accuracy classification method of garlic area.4)According to the experimental data of three years,the best recognition scheme for Qixian County garlic area recognition is to realize high accuracy of garlic area recognition based on the high score 1 WFV image in January and the support vector machine as the supervised classifier.The garlic area of 2015,2016 and 2017 was 498775 mu,530785 and 723009 Mu respectively.The recognition accuracy of garlic area was 96.2%,94.2% and 96.71% respectively,and have a good recognition accuracy.