**2. Low altitude remote sensing (LARS) system**

Site-specific management of inputs characteristic of PA promotes conservation of agricultural resources while maintaining crop viability. However, the application of satellite-based images still cannot fulfill the specific requirements of PA technology. Stafford (2000) observed that satellite images for application of PA are handicapped in terms of spectral and temporal resolution and can be affected by variable weather conditions. Lamb and Brown (2001) indicated that the low-resolution satellite images only beneficial for large-scale studies, are not appropriate for the small-scale farms prevalent in many areas of Asia, for example. Additionally, satellites providing higher-resolution images, e.g., QuickBird (DigitalGlobe, Longmont, Colo.) and ASTER (National Aeronautics and Space Administration, Washington, D.C.), have long revisit times, making them of limited utility for any application that might require frequent images (nutrient stress monitoring, for example). In the past, researchers had used manned aerial vehicles (helicopters and aero planes) to acquire surface images (Table 1). Though a large area can be mapped within short time, cost involved in the aerial vehicles is very high and also requires sophisticated system, trained operators, and professionals. Therefore, an unmanned helicopter is used for aerial image acquisition.

advantages. Okamoto et al. (2007) used a hyperspectral line-scanning camera for weed detection. This system produced hyperspectral images from a Specim ImSpector V9 imaging spectrograph mounted on a tractor that was set to move slowly through the field. Principal spectral components could be extracted and analyzed using various discrimination schemes. However, on-the-go hyperspectral sensing is slow and impractical, since enough area must

Biomass is an important trait in functional ecology and growth analysis. The typical methods for measuring biomass are destructive, laborious and time consuming. Thus, they do not allow the development of individual plants to be followed and require many individuals to be cultivated for repeated measurements. Non-destructive method may be an option to overcome these limitations. Crop residue estimation has been accomplished using RADARSAT images (Jensen et al., 1990; McNairn et al., 1998), using LANDSAT images (Thoma et al., 2004), and

Prediction of yield using remote sensing images has been practiced by many researchers (Fablo and Felix, 2001; Alvaro et al., 2007). Rice crop area has been estimated from Landsat images (Tennakoon et al., 1992) for wide-scale yield prediction. Canopy reflectance was estimated at panicle initiation stage using a portable spectroradiometer (LI-1800, LICOR) with a remote cosine receptor attached to a 1.5 m extension arm for smaller-scale yield prediction (Chang et al., 2005). Yield prediction has also been accomplished for corn (Chang et al., 2003; Kahabka et al., 2004), cotton (Thomasson et al., 2000) wheat (Doraiswamy et al., 2003), citrus (Zaman et al., 2006) and wild blueberry (Zaman et al., 2010). Tea leaf yield was estimated using vegetation indices such as normalized difference vegetation index (NDVI)

Rice (*Oriza sativa* L.), which is the staple food of most Asian countries, accounts for more than 40% of caloric consumption worldwide (IRRI, 2006). Annual rice production was approximately 590 million tons and yield was 4.21 ton ha-1 in Asia for 2006 (FAOSTAT, 2007). The profit from cultivating a rice crop is derived from the crop grain yield and total biomass produced. Predicting rice yield at or around the panicle initiation stage would

Site-specific management of inputs characteristic of PA promotes conservation of agricultural resources while maintaining crop viability. However, the application of satellite-based images still cannot fulfill the specific requirements of PA technology. Stafford (2000) observed that satellite images for application of PA are handicapped in terms of spectral and temporal resolution and can be affected by variable weather conditions. Lamb and Brown (2001) indicated that the low-resolution satellite images only beneficial for large-scale studies, are not appropriate for the small-scale farms prevalent in many areas of Asia, for example. Additionally, satellites providing higher-resolution images, e.g., QuickBird (DigitalGlobe, Longmont, Colo.) and ASTER (National Aeronautics and Space Administration, Washington, D.C.), have long revisit times, making them of limited utility for any application that might require frequent images (nutrient stress monitoring, for example). In the past, researchers had used manned aerial vehicles (helicopters and aero planes) to acquire surface images (Table 1). Though a large area can be mapped within short time, cost involved in the aerial vehicles is very high and also requires sophisticated system, trained operators, and professionals.

be covered per sweep for timely data acquisition over large field areas.

using images captured by radio-controlled model aircraft (Hunt et al., 2005).

and triangular vegetation index (TVI) (Rama Rao et al., 2007).

**2. Low altitude remote sensing (LARS) system** 

provide valuable information for future planning and yield expectations.

Therefore, an unmanned helicopter is used for aerial image acquisition.


Table 1. Comparative benefits of remote image acquisition platforms

LARS is a relatively new concept of remote image acquisition currently discussed by agriculturists implementing precision agriculture technology. As the name suggests, it is a system of acquiring images of the earth surface from a lower altitude as compared to the commercial remote sensing satellites. In this system, the images are acquired mostly below cloud cover and very near field features of interest. Low-altitude remote sensing using unmanned aerial vehicles can be an inexpensive and practical substitute for sophisticated satellite and general aviation aircraft, and it is immediately accessible as a tool for the farmer.

Various unmanned LARS systems have been developed and used in the remote image acquisition for PA applications. Some LARS platforms, kites (Aber et al., 2002), balloons (Amoroso and Arrowsmith, 2000; Seang and Mund, 2006), high-clearance tractors (Bausch and Delgado, 2005), and unmanned airplanes and helicopters (Sugiura et al., 2002; Fukagawa et al., 2003; Eisenbiss, 2004; Herwitz et al., 2004; Sugiura et al., 2004; Hunt et al., 2005; MacArthur et al., 2005, 2006; Xiang and Tian, 2006, 2007a, 2007b; Huang et al., 2008) have been successfully using for PA applications in different cropping systems.

These platforms were mounted with image acquisition devices and location measuring receivers, which can fly over agriculture farms and targeted areas for capturing images. As indicated by Sugiura et al. (2002) the major drawbacks of unmanned helicopters are limited payload capacity and precise control over working speed of the system. Thus, mounted systems operation has to be programmed properly to neutralize the effect of wind speed. The low payload capacity of the system was adjusted by selecting light weight mounting equipment and tools. Stombaugh et al. (2003) suggested replacing heavy weight professional digital cameras with light weight, low cost, commercial digital cameras. As the individual images acquired by the LARS system covers small area, geo-referenced images can be mosaic for mapping entire farmland and targeted areas. Global positioning system (GPS) was used in aerial platforms for obtaining aircraft location information (Hayward et al., 1998), for geo-referenced videobased remote sensing images (Thomoson et al., 2002) and in VRT system guidance (Fadel, 2004). Buick (2002) proposed the guidelines to select proper GPS receivers for specific applications.

Rice Crop Monitoring with Unmanned Helicopter Remote Sensing Images 257

(a) (b)

(c) Fig. 1. LARS system operation: (a) R/C helicopter mounted with image acquisition system,

> 3 (green, red, and NIR); band centers and bandwidths are fundamentally equivalent to Landsat bands TM2, TM3, and

(b) acquiring image in rice crop, and c) major components.

Image size (resolution) 1280 1024 (1.3 Mpixel)

Ground pixel resolution 0.000707 m/pixel (estimated)

TM4

Triggering Manual/cable switch triggering

Table 2. Specification of the Tetracam ADC Green-Red-NIR sensors

Characteristics Values

Spectral bands

Pixel size 6.01 micron

Lens type C-mounted

Lens 8.5 mm

Thomson and Sullivan (2006) observed that both agricultural aircraft and unmanned aerial vehicles (UAVs) may potentially more easily scheduled and accessible remote sensing platforms than the remote sensing satellites and general aviation aircraft customarily used in the U.S. However, use of agricultural aircraft is limited to those areas where aerial crop spraying is prevalent. Hunt et al. (2005) used a radio-controlled helicopter-mounted image acquisition system to estimate biomass and nitrogen status for corn, alfalfa, and soybean crops. Digital photographs have been used for site-specific weed control for grassland swards (Gebhardt et al., 2006; Beerwinkle, 2001), for tomato (Zhang et al., 2005) and for wild blueberry (Chang et al., 2011). Chen et al. (2003) using an high-elevation tractor system, indicated that multi-spectral images at 555, 660, and 680 nm wavelength band centers demonstrated good prediction ability for determining the nitrogen content of rice plants.

This chapter is intended to focus on the effectiveness of low-altitude remote sensing (LARS) images obtained by a multispectral imaging platform mounted in a radio-controlled unmanned helicopter to estimate rice crop parameters as a function of varying nutrient availability. Non-destructive image analysis technique is used to estimate rice yield and total biomass. It also examines the effectiveness of near-real time estimation of protein content from nutrient availability with rice leaf. Consistent with the fact that most multispectral cameras small enough to be used in unmanned aerial vehicles utilize predefined wavebands for feature detection, applicability of the widely used NDVI incorporating these wavebands is evaluated.

#### **2.1 System components**

A radio-controlled model helicopter (X-Cell Fury 91, Miniature Aircraft, Orlando, USA) is equipped with a Tetracam agricultural digital camera (ADC) (Tetracam, Inc., Chatsworth, Cal.), (Table 2). It is also equipped with 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 measure pressure variation for altitude measurement), COM-1288 GPS receiver (to provide position information: latitude and longitude), digital camera (to acquire multispectral (*G-R-NIR*) images) etc., monitored by a PC-104 based CPU-1232 microprocessor. A PC-104 compatible Power Supply Unit (ACS-5150), being powered from an external 12Vdc battery, is used to supply the necessary power to all the sensors including microprocessor (Figures 1 and 2).

The camera is a wideband multispectral camera utilizing a CMOS CCD (charge-coupled device) with a Bayer filter mask for multispectral imaging (Table 2). The unmanned helicopter weighed about 6 kg with a payload capacity of 5 kg. The radio console is capable of controlling the unmanned helicopter within a 1 km radius. The system uses a batteryinitiated glow fuel (250 mL) engine, supporting 15 min of flight at length. A spectroradiometer with wavelength range of 350 to 2350 nm (Spectra Co-op, Inc., Tokyo, Japan) can be used to estimate reflectance at ground level in the red (at 660 nm) and NIR bands (at 800 nm). Bandwidth at each center is 2.5 nm.

A control program, developed in "C" programming language, was used for the DOS operating system based microprocessor, to coordinate, the simultaneous clicking of digital camera and obtaining the readings from the sensors, and to store the information as a file in the storing device. The program enabled the system to acquire image and sensor reading at minimum time interval of 12 seconds. The images and corresponding sensor readings as digital number (0-255) were supplied to the image processing algorithm.

Thomson and Sullivan (2006) observed that both agricultural aircraft and unmanned aerial vehicles (UAVs) may potentially more easily scheduled and accessible remote sensing platforms than the remote sensing satellites and general aviation aircraft customarily used in the U.S. However, use of agricultural aircraft is limited to those areas where aerial crop spraying is prevalent. Hunt et al. (2005) used a radio-controlled helicopter-mounted image acquisition system to estimate biomass and nitrogen status for corn, alfalfa, and soybean crops. Digital photographs have been used for site-specific weed control for grassland swards (Gebhardt et al., 2006; Beerwinkle, 2001), for tomato (Zhang et al., 2005) and for wild blueberry (Chang et al., 2011). Chen et al. (2003) using an high-elevation tractor system, indicated that multi-spectral images at 555, 660, and 680 nm wavelength band centers demonstrated good prediction ability for determining the nitrogen content of rice plants. This chapter is intended to focus on the effectiveness of low-altitude remote sensing (LARS) images obtained by a multispectral imaging platform mounted in a radio-controlled unmanned helicopter to estimate rice crop parameters as a function of varying nutrient availability. Non-destructive image analysis technique is used to estimate rice yield and total biomass. It also examines the effectiveness of near-real time estimation of protein content from nutrient availability with rice leaf. Consistent with the fact that most multispectral cameras small enough to be used in unmanned aerial vehicles utilize predefined wavebands for feature detection, applicability of the widely used NDVI

A radio-controlled model helicopter (X-Cell Fury 91, Miniature Aircraft, Orlando, USA) is equipped with a Tetracam agricultural digital camera (ADC) (Tetracam, Inc., Chatsworth, Cal.), (Table 2). It is also equipped with 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 measure pressure variation for altitude measurement), COM-1288 GPS receiver (to provide position information: latitude and longitude), digital camera (to acquire multispectral (*G-R-NIR*) images) etc., monitored by a PC-104 based CPU-1232 microprocessor. A PC-104 compatible Power Supply Unit (ACS-5150), being powered from an external 12Vdc battery, is used to supply the necessary power

The camera is a wideband multispectral camera utilizing a CMOS CCD (charge-coupled device) with a Bayer filter mask for multispectral imaging (Table 2). The unmanned helicopter weighed about 6 kg with a payload capacity of 5 kg. The radio console is capable of controlling the unmanned helicopter within a 1 km radius. The system uses a batteryinitiated glow fuel (250 mL) engine, supporting 15 min of flight at length. A spectroradiometer with wavelength range of 350 to 2350 nm (Spectra Co-op, Inc., Tokyo, Japan) can be used to estimate reflectance at ground level in the red (at 660 nm) and NIR

A control program, developed in "C" programming language, was used for the DOS operating system based microprocessor, to coordinate, the simultaneous clicking of digital camera and obtaining the readings from the sensors, and to store the information as a file in the storing device. The program enabled the system to acquire image and sensor reading at minimum time interval of 12 seconds. The images and corresponding sensor readings as

incorporating these wavebands is evaluated.

to all the sensors including microprocessor (Figures 1 and 2).

bands (at 800 nm). Bandwidth at each center is 2.5 nm.

digital number (0-255) were supplied to the image processing algorithm.

**2.1 System components** 

(c)

Fig. 1. LARS system operation: (a) R/C helicopter mounted with image acquisition system, (b) acquiring image in rice crop, and c) major components.


Table 2. Specification of the Tetracam ADC Green-Red-NIR sensors

Rice Crop Monitoring with Unmanned Helicopter Remote Sensing Images 259

The altitude has to be selected considering the camera's field of view to acquire a single image for each treatment plot. Images with effective dimensions of 18 m 14 m were collected from a 20 m flying height, covering a single plot. For field application the height can be varied as per the suitability of the researcher, to cover wider area in each image. Flight altitude was recorded with a height sensor (MPXAZ4115A barometric sensor, Freescale Semiconductor, Austin, Tex.) mounted on the LARS system. Images are obtained at five different heights, and the images obtained closest to the 20 m height were selected for analysis. Five ground-based reflectance readings were obtained for the rice canopy and BaSO4 standard white reference board using the *Spectroradiometer* in each of the experimental plots. The ground-based readings were obtained immediately after the LARS system-based image acquisition. The plot-wise ground-based reflectance value is calculated

Multispectral images acquired by the Tetracam ADC camera (.dcm format) were converted into *.tiff* format for analysis. The *.tiff* format reduces the storage space and also effectively retains the image quality for image processing. Images were uploaded to Pixelwrench software (Tetracam, Inc., Chatsworth, Cal.), which contains programs for deriving one of several vegetation indices (.hdr format) from raw image data. An NDVI image was produced for each test plot, and the average NDVI index was estimated using the customdeveloped program in the C programming language from images acquired by the LARSmounted sensors (Figure 3). Ground-based reflectance data were collected to estimate mean NDVI of the experimental plots (NDVISPECTRO). NDVISPECTRO was estimated using the software provided by the *Spectroradiometer* manufacturer. Linear regression models can be

developed in SAS (ver. 9.1, SAS Institute, Inc., Cary, N.C.) or any standard software.

(a) (b) (c)

The normalized difference vegetation index (NDVI) is the mostly adopted reflectance index

*NIR R NDVI*

Where, *NIR*: Radiance value for *Near-infrared* band; *R*: Radiance value for *Red* band.

*NIR R* 

(1)

Fig. 3. Stages of image processing: (a) raw image with plot boundaries (as taken by the image acquisition system), (b) plot-scale image of the rice crop, and (c) NDVI image

for agricultural cropping and vegetation studies (Rouse et al., 1973) given as;

as the mean of the five readings.

**3. Validation of LARS setup** 

**2.3 Image pre-processing** 

Fig. 2. Schematic representation of the LARS image acquisition system

As individual images, of digital camera covers very small ground area, is mosaic with the algorithm to develop a single map of the study area. HIPSC software converts the digital numbers into relevant sensor readings and used them to carry out image processing operations, such as: image rotation, image mosaic and reflectance index (NDVI, Green NDVI etc.) estimation. The software can develop site-specific zone maps based on variation in reflectance index values and also provide ground control points (GCPs) for mosaic image geo-registration using commercial software.
