Unmanned Aerial Vehicle-based Methods And Techniques For Evaluating Residual Film Pollution In The Field

The plastic film mulching technology has the functions of increasing temperature,entropy,water saving and increasing production.It is widely used in agricultural production,but the residual film remained after crop harvesting is not cleared in time,which not only causes serious environmental pollution,but also reduces soil moisture content,leading to soil compaction,decreased fertility,and ultimately reduced crop yield.Therefore,the rapid and accurate assessment of the distribution of residual film pollution in the field has great practical significance for guiding the accurate residual film collecting in the field.In view of the shortcomings of traditional evaluation methods such as high labor intensity and low efficiency,the paper innovatively proposed a method for estimating residual film pollution in the field based on unmanned aerial vehicle(UAV).The paper studied the distribution characteristics of residual film in the field,and then carried out the residual film image recognition research.Combination the residual film distribution model with the image recognition results,the mathematical model of residual film distribution based on UVA was established.The accuracy of the model was verified through field experiments.The main research contents are as follows:(1)The color information and characteristics of the residual film image in the field were studied.A residual image acquisition system based on UAV was built.The residual film images of tobacco fields in different periods were collected.The low-level processing method of residual film images was analyzed.The residual image was corrected by gray interpolation method.The filtering effect of different filtering methods on the image showed that the 3×3 median filtering effect was the best.(2)Research on image segmentation algorithm of residual film based on UVA.In order to avoid the influence of illumination on the recognition accuracy of residual film,the RGB and HSV color components of the residual film and soil in the direct sunlight area and the shadow area were statistically analyzed and compared.It was found that the shadow area ofresidual filmis between direct sunlight and the shadow area ofsoilfor the term of B component while the direct sunlight and shadow areas of residual film is lower than the background for the term of S component.The manual threshold method,the iterative threshold method,the maximum inter-class variance method,the maximum entropy method,the K-mean mean clustering method and the pulse-based neural network method were used to segment theresidual film from background for both ofthe B and S components respectively.It is found that the B component was able to recognised residual filmin direct sunlight area while was not able to recognised residual filmin shadow areafrom the background;The S component was able to recognised residual filmin direct sunlight and shadow area from the background.Moreover the pulse-based neural network method based on S component had better segmentation effect,and the maximum inter-class variance method and the iterative threshold method was second.According to the sunlight direction and different crop growth periods,different recognition algorithm for identifying residual film in the field were established,and the average recognition rate was 87.42%.(3)The distribution characteristics of residual film pollution in the field were studied.The current situation of residual film pollution in typical tobacco growing areas of Bijie City was investigated by random sampling and plum blossom point sampling methods.The distribution characteristics of residual film pollution were obtained:With the increasing of film mulching years,the amount of residual film in the soil layer of 0-10 cm was the most significant,and the quality of residual film in the soil layer of 10-25 cm increased steadily;the average number and quality of residual film was 48.72 piece/m~2 and 18.549 g/m~2 respectively.The number of residual film was 10.38 piece/m~2,26.1 piece/m~2 and 12.24 piece/m~2 and the quality of residual film was 11.34 g/m~2,5.57 g/m2 and 1.639 g/m~2 respectivelyfor the three soil layersfrom the top to bottom.The longer the film was covered and the deeper the soil layer was,the smaller the fragmentation of residual film was.The fragmentation of residual film was 1.567,0.946,0.430,0.348,0.332 and 0.237 corresponding to mulching year from 1 to 6 years respectively.(4)The distribution mathematical model of the residual film pollution in the field was studied.Using Origin software to analyze the residual film distribution data,it was found that the degree of fitting,the residual film quality and quantity to the film age,was higher than 97%;At the condition of the same film covering year,the degree of fittness,the total mass of the film to the surface residual film quality,was higher than 85%.According to the fitting function,the mathematical model of residual film pollution distribution was constructed.Furthermore,the residual film pollution level of tobacco fields in Bijie City was established.Less than 6 g/m~2,6~13g/m~2,13~18 g/m~2,18~23 g/m~2,23~28 g/m~2,and more than 28 g/m~2 are regarded as level of 1,2,3,4,5,and 6 pollution degree respectively.(5)Method research for evaluating residual film pollution based on image information.A calculation method and mathematical model for residual film distribution based on image information of UAV were proposed by adding the weight value to the model.The mathematical model was validated by field experiment.The evaluation level based on UAV image information for the residual film was built.This paper provides a rapid and accurate assessment method for residual film distribution in the field.