Classification of guppies’ (Poecilia reticulata) gender by computer vision

Labor costs of guppy growers and breeders are largely those of manual sorting (by strain, quality and gender) and counting fish. In most farms, female and male fish are grown together and sold either separately or together. Sorting fish according to gender is important for marketing as well as for breeding programs, so that a device for sorting and counting fish can potentially reduce production costs and improve quality.

A project aiming to develop sorting and counting technologies for ornamental fish growers included development and testing of image-processing algorithms for sorting guppy fish (Poecilia reticulata) by gender. The algorithms are derived from shape and color differences between female and male guppies. An algorithm for the determination of landmarks on fish contours was developed and found to be accurate in accordance with human judgment, enabling extraction of specific shape and color features of the tail and the body.

The algorithms were applied to three sets of images of guppies of the “Red-Blond” strain. Gender identification accuracy was approximately 90% using shape features, approximately 96% using color features and was slightly improved when both color and shape features were used.

Some of the components used are essential for future development of a computer vision based system for sorting and grading ornamental fish by strain and quality.     

Guidance of groups of guppies (Poecilia reticulata) to allow sorting by computer vision

Prior to marketing, ornamental tropical fish are currently sorted, graded and counted manually; a process that is labor intensive, stressful to fish and inaccurate. Manipulation of fish movements at our will is a key element in the development of a device for automatically sorting, grading and counting fish by a computer vision system, and the present study is one step towards the development of such a fish-friendly sorting device. The positive phototactic and rheotactic innate responses of guppies (Poecilia reticulata) were exploited to induce them to swim through transparent pipes or narrow channels, to render them ready for inspection by a computer vision system. Water flow direction and velocity, water level, and illumination were effective in inducing fish to move from one container into the other via a narrow transparent pipe. It was attempted to separate guppies moving in groups into individually spaced fish, by introducing an obstacle into a narrow channel or narrowing the transparent pipe.     

Use of aerial thermal imaging to estimate water status of palm trees

A methodology to estimate water status of palm trees from aerial thermal images was developed. Deficit irrigation of 80% in three drip-irrigated date-palm plots in the northern Dead Sea region was manipulated during the winter of 2007 and 2008. An uncooled thermal camera was used for extensive aerial imaging to detect palm trees and pure-canopy pixels by using only aerial thermal images. An automatic procedure, based on watershed segmentation analysis, was developed which enabled detection of all palm trees in the thermal images. Two new methods were developed to select palm trees and pure pixels within them: basin-based and pixel-based. From the temperatures of pure-canopy pixels, significant differences were found between palm trees under commercial and deficit irrigation regimes, in all three plots. Automated detection of canopy, based on aerial thermal images, is a key step towards commercial mapping of within-plot water-status variability. A protocol, based on the developed methodology, was suggested for mapping water status variability in a palm plot, and for irrigation scheduling.     

Low and high-level visual feature-based apple detection from multi-modal images

Automated harvesting requires accurate detection and recognition of the fruit within a tree canopy in real-time in uncontrolled environments. However, occlusion, variable illumination, variable appearance and texture make this task a complex challenge. Our research discusses the development of a machine vision system, capable of recognizing occluded green apples within a tree canopy. This involves the detection of “green” apples within scenes of “green leaves”, shadow patterns, branches and other objects found in natural tree canopies. The system uses both thermal infra-red and color image modalities in order to achieve improved performance. Maximization of mutual information is used to find the optimal registration parameters between images from the two modalities. We use two approaches for apple detection based on low and high-level visual features. High-level features are global attributes captured by image processing operations, while low-level features are strong responses to primitive parts-based filters (such as Haar wavelets). These features are then applied separately to color and thermal infra-red images to detect apples from the background. These two approaches are compared and it is shown that the low-level feature-based approach is superior (74% recognition accuracy) over the high-level visual feature approach (53.16% recognition accuracy). Finally, a voting scheme is used to improve the detection results, which drops the false alarms with little effect on the recognition rate. The resulting classifiers acting independently can partially recognize the on-tree apples, however, when combined the recognition accuracy is increased.     

Crop water stress mapping for site-specific irrigation by thermal imagery and artificial reference surfaces

Variable-rate irrigation by machines or solid set systems has become technically feasible, however mapping crop water status is necessary to match irrigation quantities to site-specific crop water demands. Remote thermal sensing can provide such maps in sufficient detail and in a timely way. In a set of aerial and ground scans at the Hula Valley, Israel, digital crop water stress maps were generated using geo-referenced high-resolution thermal imagery and artificial reference surfaces. Canopy-related pixels were separated from those of the soil by upper and lower thresholds related to air temperature, and canopy temperatures were calculated from the coldest 33% of the pixel histogram. Artificial surfaces that had been wetted provided reference temperatures for the crop water stress index (CWSI) normalized to ambient conditions. Leaf water potentials of cotton were related linearly to CWSI values with R ² = 0.816. Maps of crop stress level generated from aerial scans of cotton, process tomatoes and peanut fields corresponded well with both ground-based observations by the farm operators and irrigation history. Numeric quantification of stress levels was provided to support decisions to divide fields into sections for spatially variable irrigation scheduling.     

Physical and visual properties of grape rachis as affected by water vapor pressure deficit

Rachis browning of table grapes after harvest is a significant problem, and water loss is considered the primary factor in browning. The major rachis desiccation and browning occurs during marketing at ambient temperatures and relative humidity (RH) which create high water vapor pressure deficits (WVPD). In this study the effect of WVPD and its components on rachis browning were examined on the two white seedless cultivars ‘Superior’ and ‘Thompson’. The grape clusters were stored at 20 °C or at 10 °C with low (70%) or high (>95%) RH, thus creating 4 WVPD levels. At each WVPD the clusters were held in open punnets, punnets sealed with low density polyethylene film or microperforated polyethylene, and examed every 2 or 3 d for weight loss, berry firmness, rachis dry weight and subjective rachis index. In addition, the rachis were photographed and image analysis employed to identify the level of browning. The results show that image analysis gave very similar patterns to subjective evaluation of rachis browning with correlation coefficients up to 0.90. However, image analysis detected an increase in browning before subjective evaluation. There was poor overall correlation between cluster weight loss and rachis dry weight to browning for ‘Superior’ grapes but a good correlation for ‘Thompson’. Rachis of ‘Superior’ suffered extensive browning at 20 °C even at high RH while rachis of ‘Thompson’ remained relatively green under similar conditions. ‘Thompson’ grape rachis remained green during the entire examination period (11 d) when held at high RH in either 10 °C or 20 °C. At high WVPD, microperforated packaging offered better control of browning in ‘Superior’ grapes than closed packaging, while clusters of ‘Thompson’ retained green rachis after 4 d in open punnets, and after 7 d in covered punnets. In summary, detailed analysis of rachis browning shows that water loss is an important but not the only factor in browning. Quantitative and objective measurement of rachis browning is likely to facilitate better communication of experimental data and higher resolution of processes which lead to browning.     

Procedure for predicting and designing moving sprinkler application patterns

Summary Water application pattern, WAP, is one of the most important factors that determine the instantaneous and the cumulative application rates of moving irrigation machines. The mathematical background of a procedure to predict and design the WAP of moving irrigation machines is introduced. It includes a mathematical analysis of the effect of pressure head, height and spacing between emitters on the WAP, and a nomograph that presents this analysis graphically and illustrates the design procedure of the application pattern of irrigation machines.Abbreviations P()a water application rate at a normalized radial distance from the emitter [m/s] - ka number of linear segments needed to represent the pattern - s/Ra normalized radial distance from the emitter - Ra wetted radius [m] - sa radial distance from the emitter [m] - n _j n _i/ha normalized water application rate at point - j, ha maximum water application rate [m/s] n _j water application rate at point j [m/s] - _j =m _j/Ra normalized radial distance of point j from emitter - m _ja radial distance of point - ja from emitter [m], CWAP - (x)a Cumulative Water Application Pattern: amount of water per unit area applied at a distance - xa from the travel path of the emitter [m³/m²] - xa distance from the travel path of the emitter [m] - T _xa time of application at a distance - xa from the travel path of the emitter [s] - va velocity of propagation of the machine [m/s] - k ₁a the outmost linear segment that its radial distance from the emitter - m _k1a is smaller than the distance of the travel path from the emitter - x, T _ja time at which the - j ^tha linear segment (ring) stops influencing the point located at a distance - xa from the emitter - ₁, ₂, ₃a dimensionless numbers derived by dimensional analysis - ua water jet velocity [m/s] - ga gravity acceleration [m/s²] - da nozzle diameter [m], v kinematic viscosity [m²/s] - Ha emitters height [m] - , a regression analysis coefficients - Paa Pattern fit coefficient for water application - F(r)a normalized desired water application pattern [1/m] - f(r)a normalized actual water application pattern [1/m] - La common distance on which - F(r) and f(r)a are defined [m], SP spacing interval between emitters [m] - DSa dimensionless spacing interval between emitters - DSa variation of dimensionless spacing interval - Paa variation of Pa coefficient - Pa pressure head [kPa]