| Abstract | In developing countries with small and medium farm holdings, mixed and diversified cropping practices, satellite-based remote sensing is found inappropriate for Precision Agriculture (PA) technology adoption, due to cost, timely availability, and low spatial and high temporal resolution of imagery. Attempts were made to develop a low cost image acquisition system mounted on an unmanned helicopter platform to provide user-specified and near-real time images for quick assessment of the soil and crop parameters. A radio controlled unmanned helicopter platform (Low-Altitude Remote Sensing -LARS) was developed to mount the image acquisition system including a multispectral camera on its landing skid, with the data acquisition system consisted of various sensors, such as: C-100 Magnetic compass (to obtain platform orientation angle from North), Inertial Measurement Unit (to obtain roll and pitch orientation angles), Barometric sensor (to estimate altitude in terms of pressure variation), COM-1288 GPS receiver (to acquire position information: Latitude and Longitude), digital camera (to acquire multispectral (G-R-NIR) images) etc. monitored by a PC-104 based CPU-1232 embedded microprocessor. A PC-104 compatible Power Supply Unit (ACS-5150), being powered from an external 12Vdc battery, supplied the necessary power to all the sensors including microprocessor and digital camera. A control program was developed in "C" programming language for the DOS operating system based microprocessor, for coordinating and sequencing the functions; the simultaneous, clicking of digital camera, acquisition of sensor readings, and storage of information as a file in the memory device. The software interface enabled the system to acquire a set of image and sensor readings at minimum time interval of 12 seconds. The images and corresponding sensor readings as digital numbers (0-255) were supplied to the HIPSC (Helicopter' Image Processing Software in "C") software for further image processing. The HIPSC software could carry out the whole image processing sequence in a single run including mosaic function to create a single map of the selected area using the individual images taken by the digital camera with necessary image rotations, convert the digital numbers into relevant sensor readings, estimation of reflectance indices NDVI, Green NDVI etc. Site-specific zone maps based on variation in reflectance index values could be generated and could use commercial software to provide ground control point (GCPs) data for mosaic image Georegistration The performance evaluation of the LARS system was carried out for rice crops in test plots of five N-treatments rates (0, 33, 66, 99 and 132 kg ha⁻¹) each having three replications, and comparing with the groundtruth readings from Spectrophotometer and SPAD 502 Chlorophyll meter estimations. Single image from each plot were acquired by the image acquisition unit of the LARS system, being operated at a height of 20m. Two set of reading, at 45 days after sowing, and 65 days after sowing (booting stage), were taken for evaluation. The coefficient of determination (r²) between N-treatments against NDVIiars, NDVIspectro, GNDVIars, and chlorophyll content estimated from leaf radiance values, were in the range from 0.70 to 0.90, showing a high-level correlation between sensor readings with variable nitrogen treatment rates. The test to verify the suitability of LARS based results against Spectrophotometer readings showed linear variation for NDVI index with r² of 0.70 and 0.80 for 45 days and 65 days old crops respectively. Similar correlation (r² =0.7) was also found for GNDVIars against NDVIspectro. Comparison of LARS based indices with Chlorophyll meter reading (for 65 days old crop) also showed strong correlation with r² >0.75, justifying the suitability of LARS images for crop status studies and nutrient management. The rice yield and total biomass for five N-treatments rates found to be significantly different at 0.05 and 0.1, respectively. The LARS system images could be used to estimate rice yield and total biomass giving high correlation with the NDVI values at booting stage, with r² of 0.95 and 0.96, respectively. The prediction of rice protein content using LARS images showed positive correlation with r² of 0.60. In the attempt made to use the LARS system to predict surface soil moisture content with G-RNIR sensor yielded weak correlation with r² of 0.34. The adoption LARS system and precision agriculture for developing countries was explained in terms of Type-I and Type-II error and recommended for quick implementation of the system for better profits |