Detection Of Nitrogen Content In Apple Leaves Based On Hue Difference

Nitrogen is one of the essential elements for the growth and development of apple trees.Obtaining timely and accurate nitrogen status of apple trees is of great significance for reasonable fertilization,production increase,and optimization of fruit quality.Nowadays most fertilization of fruit trees is based on empirical fertilization.The irrational use of nitrogen fertilizer will not only cause soil compaction and water pollution,but also cause waste of resources and increase production costs.At the same time,it will also cause a decline in fruit yield and fruit quality.However,the existing chemical analysis methods for determining the nitrogen content of apple leaves have a relatively high requirement for equipment and are difficult to popularize.At present,the application of digital image processing technology to detect nitrogen content in leaves in the visible light spectrum is concentrated in field crops,while the detection of nitrogen content in fruit trees is based on hyperspectral images and is mainly concentrated in a certain growth period and lacks for the entire growth cycle.The application of hyperspectral image analysis requires dedicated hyperspectral equipment and analysis software,which is costly and complicated.In view of the above deficiencies,this paper uses a digital camera to collect photos of leaves of three important growth stages of flowering period,mid-term and harvest period of apple trees,and studies the relationship between leaf image characteristics and leaf nitrogen content in the visible light spectral range.The main research work and innovations are as follows:(1)The relationship between the characteristic hue differences of digital image and leaf nitrogen content in apple tree leaves was studied.The research of nitrogen content detection for the leaves in the visible light spectral range of digital image processing technology mainly focuses on field crops.However,the detection of nitrogen content in apple leaves is mainly based on the problem of hyperspectral images.This article uses a digital camera to collect the photos of the leaves in the visible light.The application of digital image processing technology to the nitrogen content of apple leaves in the spectral range.(2)The relationships between leaf hue difference and leaf nitrogen content in the visible light spectral range under different nutrient conditions were studied during the three important periods of flowering stage,fruiting stage and harvest stage of fruit tree growth and development.The existing research has mainly focused on the growth and development of apple trees,and lacked the problem of detecting the nutritional status of the entire growth cycle.(3)The relationship between image characteristics and leaf nitrogen content under different light conditions was studied.Aiming at the effect of light on the image of the blade,this paper proposes a method of using the color constancy theory MSR algorithm to de-light the blade.This method can significantly improve the detection.(4)The correlation between the color difference between leaf and Pantone card and leaf nitrogen content was studied.The judgment of color contains certain subjective criteria,especially when the color difference is weak,which requires a uniform color standard to calibrate the color.This paper proposes the application of Pantone color card to calibrate the leaf green and study the leaf and Pantone color.The relationship between the difference in color tone between the cards and the nitrogen content of the leaf,and the significance test of the correlation.According to experimental verification,the nitrogen content detection model established based on the relationship between the hue difference of the digital image features and the leaf nitrogen content has a good predictability for the nitrogen content of the leaf,especially the image of the apple tree leaves after applying the MSR algorithm to remove light effects.It has a significant improvement in the prediction accuracy of the model.The correlation coefficients of the traditional flowering stage,the middle stage traditional fertilization group,the compound fertilization group,and the traditional fertilization stage were all above 0.8,0.832,0.826,0.813,and 0.817,respectively,and the correlation was the strongest.All correlation coefficients were Can pass the significance test.Each experimental group took the remaining 15 groups of data to verify the regression model.The verification results showed that the relative error between the predicted value and the actual value obtained by using the established prediction model was less than 12%.When the correlation coefficient was greater than 0.8,the model prediction was the highest.The relative error is less than 10% at this time.It is feasible to use a digital camera to collect images of apple trees in the visible spectrum,and then use digital image processing techniques to analyze the tonality of the images of the leaves to determine the nitrogen content of the apple leaves.This method is suitable for real-time and accurate monitoring of apple trees.The change of nitrogen content in leaves during the whole growth process and the rational application of nitrogen fertilizer provided a scientific theoretical basis.