A Digital Sensor with Non-Imaging Multi-Spectral and Image Modules for Continuous Monitoring of Plant Growth Conditions: Development and Validation
Xianfeng Zhou,Jingcheng Zhang,Kaihua Wu,Dongmei Chen,Huiqing Ma,Wenjiang Huang,Weiping Kong,Kaiqiang Li,Yufeng Huang,Pengtao Shi
DOI: https://doi.org/10.1016/j.compag.2024.109299
IF: 8.3
2024-01-01
Computers and Electronics in Agriculture
Abstract:In the field of precision agriculture, crop growth monitoring is integral for acquiring a holistic understanding of plant growth status, facilitating the implementation of precise and timely agronomic interventions. While traditional methods rely on labor-intensive sampling surveys, the advent of remote sensing technologies, characterized by multi- or hyperspectral imaging systems, has provided a leap forward in the non-destructive assessment of crop growth parameters. Restrictions due to technical constraints, however, curtail the widespread application of these spectral imaging modalities for plant growth surveillance. To address this challenge, this study endeavored to conceptualize and contrive a cost-effective, ground-based digital apparatus. This instrument, amalgamating a non-imaging multi-spectral module with bandwidths of 10 nm across six distinct wavelengths (530 nm, 570 nm, 680 nm, 700 nm, 740 nm, and 780 nm), a digital camera module, an operational control unit, and a solar cell assembly, was designed for continuous monitoring of crop growth. This was accomplished through intricate hardware engineering and sophisticated system control programming. Performance of multi-spectral module from the designed device were comprehensively evaluated. Results showed that measurements procured from the multi-spectral module were in strong linear correlation with irradiance observations from a canonical irradiance meter (ST-85) across all spectral bands under diverse indoor lighting scenarios, registering non-linear error (NLE) values below 3.1 %, and return error (RE) indices not exceeding 0.7 %. Continuous monitoring under variable outdoor ambient conditions-sunny and cloudy-underscored the multi-spectral module's commendable performance. Notably, in sunlit conditions, results were particularly promising: statistical metrics of relative error for relative reflectance of each wavelength, alongside vegetation indices (VIs), were detained beneath the threshold of 5 % and 3 %, respectively. Further, a rigorous cross- radiometric calibration of the multi-spectral module, benchmarked against an ASD Field Spec 4 spectrometer in tandem with paddy rice canopy, yielded precise results. The calibration models were characterized by relative root mean square error (RRMSE) values below 6 % and determination coefficients (R2) approximating 0.8 across all wavebands. Exploiting these calibrated models, Normalized difference red-edge index (ND705), 705 ), deciphered from cross-calibrated reflectance data, delivered the highest accuracy, with an RMSE of 3.76 mu g/cm2, 2 , and a RRMSE of 8.16 %, for LCC estimation in paddy rice among all the selected vegetation indices. Moreover, a lightweight model, based on combination of digital image and non-imaging multi-spectral data, achieved an overall accuracy of 95.12 % for paddy rice pests and diseases classification. These results indicated that the newly-developed device could be used for crop status monitoring.