Thomas Burks - Academia.edu (original) (raw)

Papers by Thomas Burks

Applied Engineering in Agriculture, 2003

The Yield Monitor Test Facility's performance under static grain flow was very good. First, the g... more The Yield Monitor Test Facility's performance under static grain flow was very good. First, the grain metering system accurately meters grain at varying flow rates to the clean grain elevator over the prescribed time interval. Second, the weigh scale system appears to be very accurate in measuring the accumulated mass flow in the system. However, grain stream dynamics, tank vibration, and load cell sensitivity confounds attempts to measure instantaneous grain flow at low rates of 1 to 4 kg/s. Third, the commercially available yield monitor flow sensor demonstrated excellent accuracy across a wide range of static flow rates from 1.3 to 21.1 kg/s. Percent differences in accumulated mass flow for the yield monitor and the weigh scale were less than 1% at calibration flow rates and approximately 3% for extreme flow rates. Instantaneous mass flow rate variation was observed to be approximately 8% at the mean calibration flow rate.

Frontiers in Plant Science

Identification and segregation of citrus fruit with diseases and peel blemishes are required to p... more Identification and segregation of citrus fruit with diseases and peel blemishes are required to preserve market value. Previously developed machine vision approaches could only distinguish cankerous from non-cankerous citrus, while this research focused on detecting eight different peel conditions on citrus fruit using hyperspectral (HSI) imagery and an AI-based classification algorithm. The objectives of this paper were: (i) selecting the five most discriminating bands among 92 using PCA, (ii) training and testing a custom convolution neural network (CNN) model for classification with the selected bands, and (iii) comparing the CNN’s performance using 5 PCA bands compared to five randomly selected bands. A hyperspectral imaging system from earlier work was used to acquire reflectance images in the spectral region from 450 to 930 nm (92 spectral bands). Ruby Red grapefruits with normal, cankerous, and 5 other common peel diseases including greasy spot, insect damage, melanose, scab,...

Three-dimensional (3D) reconstruction of the plant or tree canopy is an important step to measure... more Three-dimensional (3D) reconstruction of the plant or tree canopy is an important step to measure canopy geometry, volume, and leaf cover density for applications in precision agriculture, robotic harvesting, or plant phenotype. In this research, binocular stereo vision was used to recover the 3D information of the canopy. A revised camera calibration method was provided to calibrate the cameras in the world coordinate system. Only two images were used to realize a dense reconstruction. These two images were firstly rectified to make sure the corresponding feature points in the left and right images were on the same horizontal line. An efficient large-scale stereo matching (ELAS) algorithm was used to find the disparity map. The plant or tree canopy was finally reconstructed based on these calibrated camera matrices and the disparity map through a triangulation method. In this research, a series of laboratory experiments were conducted to validate the 3D reconstruction and verify th...

In the 1990s, Florida had 845,000 acres of citrus and was competitive with Brazil. That number ha... more In the 1990s, Florida had 845,000 acres of citrus and was competitive with Brazil. That number has since reduced to approximately 531,500 acres due to hurricanes, canker eradication program, urban development, economic downturn, and finally the discovery and spread of Huanglongbing (HLB), which causes tree decline and death. Many of the factors affecting the Florida citrus industry also concern other citrus producing states—such as Texas, Arizona, and California. The national threat of HLB has set the stage for developing new approaches and technologies for citrus production and harvesting that secure a future means to thrive in the midst of various invasive diseases and pests. One approach being considered is Advanced Citrus Production and Harvesting Systems (ACPHS), which uses high density semi-dwarfed trees, and intensive fertigation with optimized nutrient and water availability that accelerates plant growth. Adopting ACPHS for citrus production could increase yield production p...

In automating agricultural tasks, the ability to reach, grasp, and/or pick biological objects can... more In automating agricultural tasks, the ability to reach, grasp, and/or pick biological objects can be realized through the use of robot manipulators. As a basis for this effort, the robotic arm's performance capabilities (dexterity, repeatability, etc.) need to be assessed before an appropriate configuration is chosen. The availability of modern computing tools can simplify this process. The objective of this

Advances in Soft Computing

In this paper we compare similarity measures used for multi-modal registration, and suggest an ap... more In this paper we compare similarity measures used for multi-modal registration, and suggest an approach that combines those measures in a way that the registration parameters are weighted according to the strength of each measure. The measures used are: (1) cross correlation normalized, (2) correlation coefficient, (3) correlation coefficient normalized, (4) the Bhattacharyya coefficient, and (5) the mutual information index. The approach is tested on fruit tree registration using multiple sensors (RGB and infra-red). The combination method finds the optimal transformation parameters for each new pair of images to be registered. The method uses a convex linear combination of weighted similarity measures in its objective function. In the future, we plan to use this methodology for an on-tree fruit recognition system in the scope of robotic fruit picking

St. Joseph: ASAE, 2001

The Yield Monitor Test Facility was used to evaluate the GreenStar® mass flow sensor in a 9600 cl... more The Yield Monitor Test Facility was used to evaluate the GreenStar® mass flow sensor in a 9600 clean grain elevator under dynamically varying inflow rates. It was found that the GreenStar® yield sensor accurately predicts (less than 4% error) accumulated mass (under dynamically varying step and ramp flow rates) within normal operating conditions of 4.2 to 16.9 kg/s (10 to 40 bu/min), and does a reasonable job of predicting flow at upper limits of 20kg/s. The yield sensor accurately followed X578 variable step flow rates while operating under 20 kg/s, but exhibited significant errors for flows above 20 kg/s. This was probably due to the fact that the GreenStar® was calibrated at 12.7 kg/s. The maximum error in accumulated mass for the variable step profile was less than 2%, while the average instantaneous flow error during steady state grain metering was estimated at 4.4% with a 3.4% standard deviation. Time averaging adjacent readings reduced the error to 3.2%. It was found that transient flow had maximum accumulated flow errors of approximately 5%, while oscillating flow had errors around 2.5%. This was due to the fact that the transient flow maintains flow at the extremity of the GreenStar® operation range (based on 12.7 kg/s calibration) for longer periods. The results from these test indicated that the GreenStar® mass flow sensor accurately predicted total accumulated mass flow across a broad range of dynamic inflow rates and demonstrated good instantaneous flow prediction. Additionally, two second time averaging of adjacent flow readings smoothed noise in the original signal with minimal loss of signal definition

Food Processing Automation Conference Proceedings, 28-29 June 2008, Providence, Rhode Island, 2008

Technologies that can efficiently identify citrus diseases would assure fruit quality and safety ... more Technologies that can efficiently identify citrus diseases would assure fruit quality and safety and minimize losses for citrus industry. This research was aimed to investigate the potential of using color texture features for detecting citrus peel diseases. A color imaging system was developed to acquire RGB images from grapefruits with normal and five common diseased peel conditions (i.e., canker, copper burn, greasy spot, melanose, and wind scar). A total of 39 image texture features were determined from the transformed hue (H), saturation (S), and intensity (I) region-of-interest images using the color co-occurrence method for each fruit sample. Algorithms for selecting useful texture features were developed based on a stepwise discriminant analysis, and 14, 9, and 11 texture features were selected for three color combinations of HSI, HS, and I, respectively. Classification models were constructed using the reduced texture feature sets through a discriminant function based on a measure of the generalized squared distance. The model using 14 selected HSI texture features achieved the best classification accuracy (96.7%), which suggested that it would be best to use a reduced hue, saturation and intensity texture feature set to differentiate citrus peel diseases. Average classification accuracy and standard deviation were 96.0% and 2.3%, respectively, for a stability test of the classification model, indicating that the model is robust for classifying new fruit samples according to their peel conditions. This research demonstrated that color imaging and texture feature analysis could be used for classifying citrus peel diseases under the controlled laboratory lighting conditions.

2013 ASABE Annual International Meeting

The authors are solely responsible for the content of this meeting presentation. The presentation... more The authors are solely responsible for the content of this meeting presentation. The presentation does not necessarily reflect the official position of the American Society of Agricultural and Biological Engineers (ASABE), and its printing and distribution does not constitute an endorsement of views which may be expressed. Meeting presentations are not subject to the formal peer review process by ASABE editorial committees; therefore, they are not to be presented as refereed publications.

Transactions of the ASAE

Color co-occurrence method (CCM) texture statistics were used as input variables for a backpropag... more Color co-occurrence method (CCM) texture statistics were used as input variables for a backpropagation (BP) neural network weed classification model. Thirty-three unique CCM texture statistic inputs were generated for 40 images per class, within a six class data set. The following six classes were studied: giant foxtail, large crabgrass, common lambsquarter, velvetleaf, ivyleaf morningglory, and clear soil surface. The texture data was used to build six different input variable models for the BP network, consisting of various combinations of hue, saturation, and intensity (HSI) color texture statistics. The study evaluated classification accuracy as a function of network topology, and training parameter selection. In addition, training cycle requirements and training repeatability were studied. The BP topology evaluation consisted of a series of tests on symmetrical two hidden-layer network, a test of constant complexity topologies, and tapered topology networks. The best symmetrical BP network achieved a 94.7% classification accuracy for a model consisting of 11 inputs, five nodes at each of the two hidden layers and six output nodes (11 × 5 × 5 × 6 BP network). A tapered topology (11 × 12 × 6 × 6 BP network) out performed all other BP topologies with an overall accuracy of 96.7% and individual class accuracies of 90.0% or higher.

Journal of Imaging

Three-dimensional (3D) reconstruction of a tree canopy is an important step in order to measure c... more Three-dimensional (3D) reconstruction of a tree canopy is an important step in order to measure canopy geometry, such as height, width, volume, and leaf cover area. In this research, binocular stereo vision was used to recover the 3D information of the canopy. Multiple images were taken from different views around the target. The Structure-from-motion (SfM) method was employed to recover the camera calibration matrix for each image, and the corresponding 3D coordinates of the feature points were calculated and used to recover the camera calibration matrix. Through this method, a sparse projective reconstruction of the target was realized. Subsequently, a ball pivoting algorithm was used to do surface modeling to realize dense reconstruction. Finally, this dense reconstruction was transformed to metric reconstruction through ground truth points which were obtained from camera calibration of binocular stereo cameras. Four experiments were completed, one for a known geometric box, and the other three were: a croton plant with big leaves and salient features, a jalapeno pepper plant with median leaves, and a lemon tree with small leaves. A whole-view reconstruction of each target was realized. The comparison of the reconstructed box's size with the real box's size shows that the 3D reconstruction is in metric reconstruction.

Transactions of the ASABE, 2009

This article discusses the development of a sensor fusion system for guiding an autonomous vehicl... more This article discusses the development of a sensor fusion system for guiding an autonomous vehicle through citrus grove alleyways. The sensor system for path finding consists of machine vision and laser radar. An inertial measurement unit (IMU) is used for detecting the tilt of the vehicle, and a speed sensor is used to find the travel speed. A fuzzy logic enhanced Kalman filter was developed to fuse the information from machine vision, laser radar, IMU, and speed sensor. The fused information is used to guide a vehicle. The algorithm was simulated and then implemented on a tractor guidance system. The guidance system's ability to navigate the vehicle at the middle of the path was first tested in a test path. Average errors of 1.9 cm at 3.1 m s-1 and 1.5 cm at 1.8 m s-1 were observed in the tests. A comparison was made between guiding the vehicle using the sensors independently and using fusion. Guidance based on sensor fusion was found to be more accurate than guidance using independent sensors. The guidance system was then tested in citrus grove alleyways, and average errors of 7.6 cm at 3.1 m s-1 and 9.1 cm at 1.8 m s-1 were observed. Visually, the navigation in the citrus grove alleyway was as good as human driving.

Agricultural Engineering International Cigr Journal, Nov 27, 2013

Computers and Electronics in Agriculture, Sep 1, 2006

Current production navigation systems for agricultural vehicles rely on GPS as the primary sensor... more Current production navigation systems for agricultural vehicles rely on GPS as the primary sensor for steering control. In citrus groves, where the tree canopy frequently blocks the satellite signals to the GPS receiver, an alternative method is required. This paper discusses the development of an autonomous guidance system for use in a citrus grove. The vehicle used for this study

Proceedings of Spie the International Society For Optical Engineering, Apr 3, 2008

Citrus canker is one of the most devastating diseases that threaten citrus crops. Technologies th... more Citrus canker is one of the most devastating diseases that threaten citrus crops. Technologies that can efficiently identify citrus canker would assure fruit quality and safety and enhance the competitiveness and profitability of the citrus industry. This research was aimed to ...

Proceedings of the 19th IFAC World Congress, 2014

In this paper, a cooperative visual servo controller is presented for autonomous citrus harvestin... more In this paper, a cooperative visual servo controller is presented for autonomous citrus harvesting. A fixed camera provides a global view of a tree canopy for the camera-in-hand, attached to the end-effector, to servo to a target fruit. The paper focuses on the development of a robust, image-based, nonlinear visual servo controller to regulate the end-effector to the fruit location in the presence of unknown fruit motion. A robust feedback term is included in the controller to compensate for the bounded fruit motion, for example, due to wind gusts and robot-tree contact. Lyapunov-based stability analysis guarantees uniformly ultimately bounded regulation of the end-effector. The presented work differs from the existing methods in that the fruit motion in the form of unknown disturbance dynamics are included in the control formulation to actively compensate for the motion without the need for high-frequency image feedback.

Transactions of the ASAE, 2000

T he environmental impact from herbicide utilization has been well-documented in recent years. As... more T he environmental impact from herbicide utilization has been well-documented in recent years. As a result, efforts to protect groundwater quality by restricting the use of certain herbicides is likely to increase in the years to come. However, society is likewise concerned about maintaining cost-effective production of agricultural crops. According to Bridges and Anderson (1992), "Weeds pose one of the most important threats to our supplies of food and fiber. Losses in both yield and quality of crops due to weeds, as well as costs of weed control, constitute an enormous problem in all agricultural areas." The reduction in use of herbicides without a viable alternative will likely result in decreased production and thus higher unit production costs. Chemical application technologies are being targeted to reduce herbicide use and maintain production levels. Selective herbicide application to weed-infested areas of the field, rather than the entire field, was suggested by Thompson et al. (1991). The goal of this research was to investigate the potential for using a machine vision system to identify localized crop areas which are threatened by weed competition, thereby enabling reduced reliance on chemical application in other areas of a field.

2013 Kansas City, Missouri, July 21 - July 24, 2013, 2013

Three dimensional (3D) reconstruction of the plant or tree canopy is an important step in order t... more Three dimensional (3D) reconstruction of the plant or tree canopy is an important step in order to measure canopy geometry, such as, height, width, volume, and leaf cover. In this research, binocular stereo vision was used to recover the 3D information of the canopy. A revised camera calibration method was provided to calibrate the cameras in world coordinate system. Only two images were used to realize a dense reconstruction. These two images were firstly rectified to make sure the corresponding feature points in the left and right images were on the same horizontal line. An efficient large scale stereo matching (ELAS) algorithm was used to find the disparity map. The plant or tree canopy was finally reconstructed based on these calibrated camera matrices and the disparity map through a triangulation method. A plant (croton) with big leaves and a small citrus tree with small leaves were used to test this two-view dense reconstruction. It was easy to measure the geometry of the big leaf. Two big leaves from croton plant were used to measure the width and length of the leaves. The measurement from the reconstruction and manual measurement showed that this reconstruction was metric reconstruction. Another three reconstructions were completed based on a side view of the croton plant, a top view of the croton plant, and a side view of the citrus tree. All these gave good 3D visualization of the objects.