S. Nor Hafizah and B. Siti KhairunnizaJ. Trop. Agric. and Fd. Sc. 39(1)(2011): 000 000
Colour spaces for paddy soil moisture content determination(Ruang warna untuk penentuan kandungan kelembapan tanah sawah)
S. Nor Hafizah* and B. Siti Khairunniza*
Keywords: soil moisture content, colour, image processing, RGB, HSV, CIELUV.
AbstractA study using RGB, HSV and CIELUV colour spaces was conducted to determine the paddy soil moisture content at two different soil depths. Results from the experiment showed that each layer of soil gave a different soil colour as the soil moisture content varies with the depth of the soil layer. By comparing the result of laboratory work with the result of image processing technique, it was shown that 15 cm depth soil which had higher moisture content gave a lower value of mean pixel intensity compared to the surface soil. When the digital images of soil were transformed to HSV (USGS Munsell), a colour space commonly used to represent soil colour, the result showed that the mean pixel intensity was not consistent for each soil layer. To overcome this problem, the RGB and CIELUV colour spaces were used. The CIELUV colour space gave more consistent mean pixel intensity for each soil layer. It successfully indicated that lower moisture content will give higher value of mean pixel intensity. Results from statistical analysis also showed that RGB and CIELUV colour spaces were significantly related to the soil moisture content. CIELUV gave the highest value of correlation at 0.548 and a smaller value of RMSE in linear regression analysis.
*Department of Biological and Agricultural Engineering, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, MalaysiaAuthors full names: Nor Hafizah Sumgap and Siti Khairunniza BejoE-mail: firstname.lastname@example.orgMalaysian Agricultural Research and Development Institute 2011
IntroductionRice is a staple food in this region and most of the paddy in Malaysia is grown on low land paddy fields, flooded with water. Water depth and late flooding of field after sowing can affect yield production. It is important to monitor the water, weeds and fertilizer status to give advance notice to farmers on the occurrence of pests and diseases. The knowledge of moisture content gives advantage in agriculture practice. It gives information of soil condition and can be used to identify the best practice to improve yield production.
Moisture content is defined as the amount of water present in the soil. Moisture conditions can affect the soil structure. Wet soil can suffer from erosion while soil that is too dry can become hard and compacted. Different types of soil respond to moisture differently. A clay soil will absorb more water while sandy soil will drain water quickly. The change in moisture content may change the rheological characteristic of soil and affect many hydraulic and transport properties. Soil moisture information is valuable to a wide range of government agencies and private companies concerned with weather and
Colour spaces for paddy soil
climate, runoff potential and flood control, soil erosion and slope failure, reservoir management, geotechnical engineering, and water quality. Soil moisture is a key variable in controlling the exchange of water and heat energy between the land surface and the atmosphere through evaporation and plant transpiration (Koster et al. 2004). There are several methods to determine moisture content. The value of moisture content can be used not only to determine appropriate irrigation time but also to determine the amount of soil moisture used by the plants. In laboratory work, soil is oven dried for 24 h to get the mass of water pore. However, this method is time consuming and the calculated gravimetric moisture content may underestimate the actual amount of moisture used by plants as it measures soil moistures based on weight. The soil moisture used by plants is expressed as a depth of water. Therefore, the volumetric moisture is more appropriate for determining the amount of soil moisture used by plants as it measures soil moisture based on volume. However, it requires more tedious work compared to gravimetric method (Blaine and Steve 1998). For in situ measurement, there are available sensors such as dielectric moisture sensor which is buried under the soil and gives fast reading. However, the use of sensors is limited since it depends on the technical specifications provided by manufacturers. Soil moisture content and other physical soil properties have a strong influence on the amount and composition of energy reflected and emitted from soil surface. This effect is mainly determined by the soil moisture content of the soil surface, which is highly dynamic, being temporally and spatially variable as a result of microclimate, evaporative demands and non-uniform soil physical properties (Huete and Warrick 1990). Colour is frequently used by soil scientists for soil identification and classification. It is used as an indicator of field soil physical, chemical and biological properties and the occurrence
of soil processes. Soils with dark surface horizons are generally associated with high organic matter contents, categorizing them as fertile and suitable for plant growth (Schulze et al. 1993). Soil colours can also be used to qualitatively describe the status of moisture, for example, due to changes in the refractive index, dry soils are lighter in colour compared to the wet soils (Thompson and Bell 1996). The Munsell Soil Colour Charts can be used to precisely describe soil colours by determining the hue, value and chroma of the colour. These three variables describe a perceptual colour space and not a quantitative measure of visible light. The system was designed to arrange colours according to the equal intervals of visual perception. Thus, the primary advantage of the Munsell system is its ease of interpretation. However, the Munsell Hue Value Chroma (HVC) coordinates are psycho sensory and based on subjective perception and comparison, therefore, the system is not uniform (Viscarra Rossel et al. 2006). Although this system is useful, it is light dependent and has a limitation on the number of colour chips. As a result, this system is less appropriate when precise colour measurements are needed. These limitations have prompted researchers to use more objective and more accurate methods to measure the colour by optical techniques such as digital cameras (Viscarra Rossel et al. 2003; Viscarra Rossel et al. 2009), chromameters (Konen et al. 2003) and spectrometers (Post et al. 1993). There are a number of colour spaces that exist to overcome some limitation of the Munsell System such as Red Green Blue (RGB), CIELAB and CIELUV. CIE stands for the International Commission on Illumination. It has been defined as a system that classifies colour according to human visual system. It measures the sensitivities of the three broad bands in the eye by matching spectral colours to specific mixtures of three coloured lights (Adrian and Alan 1998). The three colours in LAB
S. Nor Hafizah and B. Siti Khairunniza
are Luminance, A (Red Green axis), and B (Blue Yellow axis). Meanwhile, the three colours in LUV are Luminance, U (Red vs Green) and V (Blue vs Yellow). CIELAB and CIELUV are device independent colour spaces and use Cartesian-type coordinate system which is more useful in numerical and predictive analysis. CIELUV has an associated two-dimensional chromaticity chart which is useful for showing additive colour mixtures, making CIELUV useful in applications using Cathode Ray Tube (CRT) displays. In order to evaluate the usefulness of the various colour space, the relationships between soil colour and soil moisture content have been analysed. Soil moisture content was correlated to parameters from the various colour spaces that describe both the lightness of the colour and its chroma. The main purpose of this study was to find appropriate colour spaces for paddy soil moisture content determination by using an image processing approach. Three different colour spaces were used to determine the soil moisture content taken from two different depths of paddy soil.
Materials and methodsSoil sampling and image acquisitionThe soil samples were collected after plowing operation from two paddy fields at Sawah Sempadan, Selangor. There were 80 samples for surface soil and 80 samples for 15 cm depth soil. The image of each sample was captured using digital camera (Canon EOS400D) and saved in a RGB colour with JPEG format. The camera was setup on a tripod (40 cm height) in a constant position and lens aperture under control lighting environment as shown in Plate 1.
Moisture content determinationGravimetric soil sampling measures soil moisture content on a weight basis by dividing the weight of water in sample by the dry weight of the sample. The American Society for Testing and Materials (ASTM) standard methods for laboratory test were
used to determine the soil moisture content. The samples were placed in a can and weighed before oven dry at 105 C for 24 h. It was then weighed again to get the mass of dry soil. The calculation for moisture content determination is as follows: MmsMd W = x 100 (1) MdMc
where, W = moisture content Mms = mass of moisture soil Md = mass of dry soil Mc = mass of can
Converting images to colour spacesThe RGB soil images were converted to HSV and CIELUV colour spaces. These images were then transformed into grayscale images. Grayscale image is an image which only carries intensity information. It is composed of shades of gray, varying from black to white. The converting process was done by using MATLAB software.
RGB to HSV conversion HSV [United States Geological Survey (USGS) Munsell] is a non-linear transformation of the R