A Novel Method for Rapid Detection of Crop Pest Status Based on NCC Combined with Edge Detection: ANAED

As computer vision and deep learning are gradually combined for high-quality target detection, higher requirements are also put forward for the embedded devices carried, which poses a huge challenge to the cost of modern agricultural precision planting and intelligent planting. Because of the above problems, this paper proposes a new method for rapid plant pest status detection based on the combination of the NCC (normalized correlation coefficient) algorithm and edge detection technology, which not only avoids the need for specific plant learning and training in deep learning but also is simple and efficient, especially suitable for low-power embedded devices such as OpenMV, RaspberryPI, etc. In practical applications, this method first uses the optimized NCC algorithm to quickly lock the observable leaves of the plant (mainly the leaves are unobstructed) and then uses edge detection technology (optimized morph-function) to extract all the edge information of the leaves, and further optimizes the edge detection results through threshold and filtering. Finally, the pest status and severity of the plant are judged according to the average value of the integrity and damage degree of the edge of the crop leaves detected per unit area. The experimental results show that this method can accurately detect the pest status of crops in a fast time, with high practicality and accuracy. In addition, through experiments on fruit-bearing crops such as cucumbers, eggplants, pumpkins, and zucchini, it was found that this method has the best detection effect in the growth and maturity stages of these plants, with an accuracy rate of about 77%-86%. This method is simple, fast, and can be applied in low-power embedded devices. It does not require deep learning and large amounts of data calculation and has high practical value.