A novel method for rapid detection of crop water shortage based on optimized NCC combined with adaptive color recognition: ANACD

In the face of the rising computing costs and dependence on high-end equipment brought by deep learning in smart agriculture applications, this paper proposes an innovative method for detecting crop water shortage status, ANACD. The algorithm design integrates an optimized Normalized Cross-Correlation (NCC) algorithm with adaptive multi-color gamut color recognition technology. This combination not only eliminates the cumbersome process of specific plant data collection, training, and recognition inherent in current deep learning architectures but also achieves one-step recognition of crop water shortage status. ANACD is perfectly adapted to mainstream low-power embedded devices such as OpenMV and Raspberry Pi due to its simplicity and efficiency. In the specific implementation, ANACD first uses the optimized NCC algorithm to accurately locate the leaves of crops and then employs the adaptive multi-color gamut color recognition algorithm to capture the color information of the leaves. Finally, the average color change of the detected crop leaf set per unit area is used to quickly and accurately judge the water shortage status of the crops. Experimental data show that ANACD can provide accurate water shortage diagnosis in a very short time, with excellent practicality and accuracy. Notably, the detection accuracy of this method remains stable at 76%-83% during the growth and maturity period of crops such as cucumbers, eggplants, pumpkins, and zucchini, meeting most agricultural application scenarios. With its advantages of simplicity, speed, and efficiency, ANACD can complete the corresponding accurate recognition tasks without relying on deep learning and complex data calculations, demonstrating significant practical potential for deployment on low-power embedded devices.