Diagnosis Of Tomato Nutrient Deficiency Based On Computer Vision Technology Under Natural Illumination

Based on referring to much literature, the research of tomato nutrient deficiency under natural illumination was put forward. The images were collected by digital camera intelligently. The main purpose of this study is to diagnose tomato disease intellectively by using plant physiology, establishment planting, digital images processing and pattern recognition. The main contents of the study are as follows:(1) Cultivated purity nutrient deficiency specimen. Deficiency of N, Ga, Mg and Fe during the growth time of tomato used as experimental objects. The ShanQi method was used for cultivating purity tomato nutrient deficiency.(2) Advanced a new method, super-green added B, for segmenting images from complex setting, which is effective by experiment. The ratio of maximal square subtract method was used to computer the threshold, which, was not been designed the parameters of man-made but variance dynamic method. Different image pre-processing ways such as removing noise and enhancing images were used on images.(3) For getting the input feature vectors of pattern recognition, the color feature and textural features extraction ways were discussed in the study. The percentage of disease leaves area to total area is a color feature parameter. The color feature includes the percentage of area and the square subtract method etc. The textural feature of individual leaves were extracted by different methods. Some were extracted by grads-gray co-occurrence and Fourierism method. Some were extracted by the wavelet method, which was especially suitable for the time-frequency region features.(4) A recognition model for deficiency leaves was built by using fuzzy K-neighbor method. Selected the effective feature parameters as the input feature vectors of pattern recognition. Using the above color and textural feature as input vectors of the recognition system, the classification accuracies of the testing data set were 92.5%, 75%, 80%, 70% and 85%, respectively, for nutrient deficiency leaves, deficiency of Ca, deficiency of Mg, deficiency of Fe, deficiency of N.The research provides important basic theories and techniques for further investigation development of plant deficiency recognition system, which was valuable in business appliance. It is of great significance for promoting the application of compute vision in agriculture engineering field.