Tiebiao Zhao

Fractional calculus in image processing: a review

Abstract Over the last decade, it has been demonstrated that many systems in science and engineering can be modeled more accurately by fractional-order than integer-order derivatives, and many methods are developed to solve the problem of fractional systems. Due to the extra free parameter order α, fractional-order based methods provide additional degree of freedom in optimization performance. Not surprisingly, many fractional-order based methods have been used in image processing field. Herein recent studies are reviewed in ten sub-fields, which include image enhancement, image denoising, image edge detection, image segmentation, image registration, image recognition, image fusion, image encryption, image compression and image restoration. In sum, it is well proved that as a fundamental mathematic tool, fractional-order derivative shows great success in image processing.

A detailed field study of direct correlations between ground truth crop water stress and normalized difference vegetation index (NDVI) from small unmanned aerial system (sUAS)

Aerial images with high spatial resolution and high temporal resolution were used to detect water stress based on canopy level normalized difference vegetation index (NDVI). We attempted to determine the correlation between stem water potential (SWP) and canopy NDVI with and without shade. Results indicated that removing the shade from the canopy improved the correlation between the NDVI of canopy and SWP with coefficient of determination (R2) from 0.001 to 0.0052. We further compared SWP and the NDVI of the canopy without shade over a period of one week to four weeks. The correlation between NDVI with SWP was highest in the time range of three weeks. However, both cases show that there is no obvious relationship between NDVI of canopy and SWP. Therefore, canopy level NDVI does not indicate water stress. Further research is needed beyond pretty pictures.

An analysis of the effect of the bidirectional reflectance distribution function on remote sensing imagery accuracy from Small Unmanned Aircraft Systems

Small Unmanned Aircraft Systems (SUASs) are increasingly being utilized for remote sensing applications due to their low-cost availability and potential for the collection of high-resolution on-demand aerial imagery. However, the field is still maturing, and there remains many questions on the accuracy and the validity of the data collected. While many researchers have investigated means of improving calibrations and data collection techniques, there are other sources of error that require investigation. In this paper, two unique characteristics of SUAS remote sensing are analyzed as potential sources of error: the use of wide field-of-view (FOV) imaging sensors and solar motion during one or more data collection flights. Both of these characteristics are related to the bidirectional reflectance distribution function (BRDF), a description of light reflection as a function of illumination direction and observer viewing angles. The wide FOV of many imaging equipment creates an inherent radial variation in viewing angle, and the solar motion creates a non-static illumination source. The results of this paper indicates that these two factors have significant contributions to errors and should not be assumed to be negligible.

Fractional order robust visual servoing control of a quadrotor UAV with larger sampling period

Unmanned aerial vehicles (UAVs) are widely applied in both civil and military fields, such as rescue, surveillance, exploration, navigation, precision agriculture and etc., because of small size, low cost and easy maintenance. However the autonomous flight of UAVs under unstructured environment is still open, especially when GPS is unavailable or indoor task is scheduled. Furthermore, GPS-based navigation accuracy needs to be improved. To deal with these limitations, in this paper, we provide an engineering-oriented solution of precise hovering based on visual servoing and fractional order proportional-integral-derivative (PID) controller without GPS information. First, the mathematical speed model of a quadcopter is obtained by step response experiments. Then, fractional order PID controller algorithm is designed to improve its hovering accuracy and robustness. In our work, the position reference is extracted with an on-board color based target recognition algorithm instead of GPS. Both simulation and field experimental results demonstrate that the proposed scheme can achieve a better performance in terms of hovering accuracy and robustness to disturbances. In particular, for the first time, we show that, the sampling period can be bigger than usual when using fractional order control algorithm, relaxing costly hardware requirement for fast real-time vision-based feedback.

Maximum Power Point Tracking of Proton Exchange Membrane Fuel Cell With Fractional Order Filter and Extremum Seeking Control

Proton Exchange Membrane FC (PEMFC) is widely recognized as a potentially renewable and green energy source based on hydrogen. Maximum power point tracking (MPPT) is one of the most important working conditions to be considered. In order to improve the searching performance such as convergence and robustness under disturbance and uncertainty, a kind of fractional order low pass filter (FOLPF) is applied for the MPPT controller design based on general Extremum Seeking Control (ESC). The controller is designed with FOLPF and high pass filter (HPF) substituting the normal LPF and HPF in the original ESC design. With this FOLPF ESC, better convergence and smooth performance is gained while maintaining the robust specifications. Simulation results are included to validate the proposed new FOLPF ESC scheme under disturbance and comparisons between FOLPF ESC and general ESC method are also provided.Copyright

Challenges in Water Stress Quantification Using Small Unmanned Aerial System (sUAS): Lessons from a Growing Season of Almond.

We conducted a study in a large almond farm in Merced County, California, to monitor water status by using high-resolution multi-spectral imagery accquired by a Small Unmanned Aerial System (sUAS). More specifically, we would like to predict Stem Water Potential (SWP) via canopy Normalized Difference Vegetation Index (NDVI). During 2014, an aircraft equipped with multi-spectral cameras flew over the orchard weekly throughout the growing season. At the same time, SWP was measured for the sample trees under five different water treatment levels. Instead of averaging pixels in an orchard level, a block level or a canopy level, pixels were analyzed in the sub-canopy level to obtain canopy NDVI. An improved correlation between SWP and canopy NDVI was obtained by applying lower NDVI threshold. The relationship between SWP and canopy NDVI was also discussed at different growing stages–fruit development and post-harvest. However, tests of equality of distribution indicated that canopy NDVI distributions from different flights within a day were significantly different. Therefore, further calibration regarding the effects of solar motion on canopy NDVI is necessary.

Maximum power point tracking with fractional order high pass filter for proton exchange membrane fuel cell

Proton exchange membrane fuel cell (PEMFC) is widely recognized as a potentially renewable and green energy source based on hydrogen. Maximum power point tracking (MPPT) is one of the most important working conditions to be considered. In order to improve the performance such as convergence and robustness under disturbance and uncertainty, a fractional order high pass filter (FOHPF) is applied for the MPPT controller design based on the traditional extremum seeking control (ESC). The controller is designed with integer-order integrator (IO-I) and low pass filter (IO-LPF) together with fractional order high pass filter (FOHPF), by substituting the normal HPF in the original ESC system. With this FOHPF ESC, better convergence and smoother performance are achieved while maintaining the robust specifications. First, tracking stability is discussed under the commensurate-order condition. Then, simulation results are included to validate the proposed new FOHPF ESC scheme under disturbance. Finally, comparison results between FOHPF ESC and the traditional ESC method are also provided.

A low cost research platform for modeling and control of multi-input multi-output fractional order dynamic systems

This paper introduces a low cost temperature control experimental platform for hardware-in-the-loop simulations. It is developed based on the thermal electronic elements to emulate a temperature control loop in a chemical process. The novel aspect lies in the discovery and discussion of some interesting characteristics of this hardware setup through experiments, which includes the fractional order inverse response behavior, the “open loop linear while closed loop nonlinear” behavior and so forth. These phenomena provide more physical evidences for revealing the value of fractional calculus.

Onion irrigation treatment inference using a low-cost hyperspectral scanner

Many studies have shown that hyperspectral measurements can help monitor crop health status, such as water stress, nutrition stress, pest stress, etc. However, applications of hyperspectral cameras or scanners are still very limited in precision agriculture. The resolution of satellite hyperspectral images is too low to provide the information in the desired scale. The resolution of either field spectrometer or aerial hyperspectral cameras is fairly high, but their cost is too high to be afforded by growers. In this study, we are interested in if the flow-cost hyperspectral scanner SCIO can serve as a crop monitoring tool to provide crop health information for decision support. In an onion test site, there were three irrigation levels and four types of soil amendment, randomly assigned to 36 plots with three replicates for each treatment combination. Each month, three onion plant samples were collected from the test site and fresh weight, dry weight, root length, shoot length etc. were measured for each plant. Meanwhile, three spectral measurements were made for each leaf of the sample plant using both a field spectrometer and a hyperspectral scanner. We applied dimension reduction methods to extract low-dimension features. Based on the data set of these features and their labels, several classifiers were built to infer the field treatment of onions. Tests on validation dataset (25 percent of the total measurements) showed that this low-cost hyperspectral scanner is a promising tool for crop water stress monitoring, though its performance is worse than the field spectrometer Apogee. The traditional field spectrometer yields the best accuracy as high as above 80%, whereas the best accuracy of SCIO is around 50%.

Comparing U-Net convolutional networks with fully convolutional networks in the performances of pomegranate tree canopy segmentation

In the last decade, technologies of unmanned aerial vehicles (UAVs) and small imaging sensors have achieved a significant improvement in terms of equipment cost, operation cost and image quality. These low-cost platforms provide flexible access to high resolution visible and multispectral images. As a result, many studies have been conducted regarding the applications in precision agriculture, such as water stress detection, nutrient status detection, yield prediction, etc. Different from traditional satellite low-resolution images, high-resolution UAVbased images allow much more freedom in image post-processing. For example, the very first procedure in post-processing is pixel classification, or image segmentation for extracting region of interest(ROI). With the very high resolution, it becomes possible to classify pixels from a UAV-based image, yet it is still a challenge to conduct pixel classification using traditional remote sensing features such as vegetation indices (VIs), especially considering various changes during the growing season such as light intensity, crop size, crop color etc. Thanks to the development of deep learning technologies, it provides a general framework to solve this problem. In this study, we proposed to use deep learning methods to conduct image segmentation. We created our data set of pomegranate trees by flying an off-shelf commercial camera at 30 meters above the ground around noon, during the whole growing season from the beginning of April to the middle of October 2017. We then trained and tested two convolutional network based methods U-Net and Mask R-CNN using this data set. Finally, we compared their performances with our dataset aerial images of pomegranate trees. [Tiebiao- add a sentence to summarize the findings and their implications to precision agriculture]