<p>Unmanned aerial vehicles (UAV) have revolutionized data collection in large scale agronomic field trials (10+ ha). Vegetative index (VI) maps derived from UAV imagery are a potential tool to characterize temporal and spatial treatment effects in a more efficient and non-destructive way compared to traditional data collection methods that require manual sampling. The overall objective of this study was to characterize and quantify maize responses to experimental treatments in field-scale research using UAV imagery. The specific objectives were: 1) to assess the performance of several VI as predictors of grain yield and to evaluate their ability to distinguish between fertilizer treatments, and the effects of removing soil and shadow background, 2) to assess the performance of VI and canopy cover fraction (CCF) as predictors of maize biomass at vegetative and reproductive growth stages under field-scale conditions, and 3) to compare the performance of VI derived from consumer-grade and multispectral sensors for predicting grain yield and identifying treatment effects. For the first objective, the results suggest that most VI were good indicators of grain yield at late vegetative and early reproductive growth stages, and that removing soil background improved the characterization of maize responses to experimental treatments. For objective two, overall, CCF was the best to predict biomass at early vegetative growth stages, while VI at reproductive growth stages. Finally, for objective three, performance of consumer-grade and multispectral derived VI were similar for predicting grain yield and identifying treatment effects.</p>

Cameras are widely used as sensors for a variety of engineering applications. In a typical video-based application, spatial segmentation is a fundamental step which provides the spatial positions of different targets for further analysis. In this thesis, we focus on videos analytics applied to the agricultural industry and describe several video segmentation methods in the context of two practical projects: autonomous farming vehicles and analysis of dairy cow health. In the autonomous farming vehicle project, we propose three spatial segmentation methods based on traditional video features to isolate the regions of the video frame where critical information appears. Two applications that apply the segmentation method are presented: farming activity classification and header-height control for a combine harvester. In the project on cow health, we propose a cow structural model based on the keypoints of joints from a side-view cow video. A detection system is developed using deep learning techniques to automatically extract the structural model from the videos. Based on this model, we also present a preliminary application which estimates the cow’s weight based on video information.<div><br></div>

<p>In this project, CNN has been applied as a training tool to process image classification and object avoidance on remote robotic cars built with the Nvidia Jetson Nano developer kit. The kit was programmed using the wireless programming environment, Jupyter notebook. In addition, two different CNN models have been applied to analyze the output result performance. The main purpose is to train the robot to identify objects and improve its accuracy. The recognition and accuracy rate under different conditions can be observed by comparing the two models with different graphic inputs conditions. This project adopts the pre-train model for real time demonstrations and can be executed in a cloudless environment (without networks involved). As a result, the robot can achieve a high accuracy rate in both CNN models output result performance. Moreover, the pre train model can execute in local service to accomplish cloudless.</p>

<p>Embedding information into a printed image is useful in many aspects, in which reliable channel encoding/decoding systems are crucial due to the information loss and error propagation during transmission. So how to improve the transmission accuracy and control the decoding error rate under a predictable level is always crucial to the channel design.</p><p>The current dissertation aims to discuss the design and performance of a two-dimensional coding method for printed materials – Circular Coding. It is a general two-dimensional coding method that allows data recovery with only a cropped portion of the code, and without the knowledge of the carrier image. While some traditional methods add redundancy bits to extend the length of the original massage length, this method embeds the message into image rows in a repeated and shifted manner with redundancy, then uses the majority votes of the redundant bits for recovery.</p><p>We introduce the encoding and decoding system and investigate the performance of the method for noisy and distorted images. For a given required decoding rate, we model the transmission error and compute the minimum requirement for the number of bit repeats.</p><p>Also, we develop a closed form solution to find the the corresponding cropped-window size that will be used for the encoding and decoding system design.</p><p>Finally, we develop a closed-form formula to predict its decoding success rate in a noisy channel under various transmission noise levels, using probabilistic modeling. The theoretical result is validated with simulations. This result enables the optimal parameter selection in the encoder and decoder system design, and decoding rate prediction with different levels of transmission error.</p><p>We also briefly discuss two other projects: development of print quality troubleshooting tools and text line detection in scanned pages.</p>

<p>Digital image processing techniques have many significant applications in industry. In this thesis, we focus on three problems in digital image processing. These three problems involve halftone images, information encoding and decoding, image alignment, and deep learning.</p><p>Specifically, the first problem is based on data-bearing halftone images, which are an aesthetically pleasing alternative to barcodes. We address the issues generated in the camera captured image alignment process. We perform some theoretical analysis and validate it by simulation. We also provide an optimal solution to the problem.</p><p>The second problem is about the alignment technique on a 3D surface. We develop a pipeline of surfaces coding to solve the alignment issues on 3D surfaces, which includes oblique surfaces and cylindrical surfaces.</p><p>The third problem is related to image retrieval. We propose a deep learning based solution to the fashion image retrieval task. Fashion image retrieval is significant to improve the customers’ experience in online shopping. A fast, accurate shopping item information retrieval system based on the customers’ uploaded image has been built by us. A novel solution is provided, and it achieves state-of-art accuracy in shopping items’ information retrieval.</p>

<div>Machine learning has been revolutionizing our approach to image understanding problems. However, due to the unique nature of the problem, finding suitable data or learning from limited data properly is a constant challenge. In this work, we focus on building machine learning pipelines for fashion and inkjet image analysis with limited data. </div><div><br></div><div>We first look into the dire issue of missing and incorrect information on online fashion marketplace. Unlike professional online fashion retailers, sellers on P2P marketplaces tend not to provide correct color categorical information, which is pivotal for fashion shopping. Therefore, to assist users to provide correct color information, we aim to build an image understanding pipeline that can extract garment region in the fashion image and match the color of the fashion item to a pre-defined color categories on the fashion marketplace. To cope with the challenges of lack of suitable data, we propose an autonomous garment color extraction system that uses both clustering and semantic segmentation algorithm to extract the identify fashion garments in the image. In addition, a psychophysical experiment is designed to collect human subjects' color naming schema, and a random forest classifier is trained to given close prediction of color label for the fashion item. Our system is able to perform pixel level segmentation on fashion product portraits and parse human body parts and various fashion categories with human presence. </div><div><br></div><div>We also develop an inkjet printing analysis pipeline using pre-trained neural network. Our pipeline is able to learn to perceive print quality, namely high frequency noise level, of the test targets, without intense training. Our research also suggests that in spite of being trained on large scale dataset for object recognition, features generated from neural networks reacts to textural component of the image without any localized features. In addition, we expand our pipeline to printer forensics, and the pipeline is able to identify the printer model by examining the inkjet dot pattern at a microscopic level. Overall, the data-driven computer vision approach presents great value and potential to improve future inkjet imaging technology, even when the data source is limited.</div>

<div> <div> <div> <p>Composite materials have become widely used in engineering applications, in order to reduce the overall weight of structures while retaining their required strength. Due to their light weight, relatively high stiffness properties, and formability into complex shapes, discontinuous fiber composites are advantageous for producing small and medium size components. However, qualifying their mechanical properties can be expensive, and therefore there is a need to improve predictive capabilities to help reduce the overall cost of large scale testing. To address this challenge, a composite material consisting of discontinuous glass fibers in a polypropylene matrix is studied at the microstructural level through coupled experiments and simulations, in order to uncover the mechanisms that cause microvoids to initiate and progress, as well as certain fiber breakage events to occur, during macroscopic tension. Specifically, this work coupled in-situ X-ray micro computed tomography (μ-CT) experiments with a finite element simulation of the exact microstructure to enable a 3D study that tracked damage initiation and propagation, and computed the local stresses and strains in the microstructure. In order to have a comprehensive 3D understanding of the evolution of the microstructure, high fidelity characterization procedures were developed and applied to the μ-CT images in order to understand the exact morphology of the microstructure. To aid in this process, ModLayer - an interactive image processing tool - was created as a MATLAB executable, and the 3D microstructural feature detection techniques were compared to traditional destructive optical microscopy techniques. For damage initiation, this work showed how high hydrostatic stresses in the matrix can be used as a metric to explain and predict the exact locations of microvoid nucleation within the composite’s microstructure. From a damage propagation standpoint, matrix cracking - a mechanism that has been notably difficult to predict because of its apparent stochastic nature - was studied during damage propagation. The analysis revealed the role of shear stress in fiber mediated flat matrix cracking, and the role of hydrostatic stress in fiber-avoidance conoidal matrix cracking. Overall, a sub-fiber simulation and an in-situ experimental analysis provided the microstructural physical phenomena that govern certain damage initiation and progression mechanisms, further enabling the strength and failure predictions of short fiber thermoplastic composites. </p></div></div></div>

No description available.

<p>Print quality (PQ) is one of the most significant issues with electrophotographic printers. There are many reasons for PQ issues, such as limitations of the electrophotographic process, faulty printer components, or other failures of the print mechanism. These reasons can produce different PQ issues, like streaks, bands, gray spots, text fading, and color fading defects. It is important to analyze the nature and causes of different print defects to more efficiently repair printers and improve the electrophotographic process. </p> <p><br></p> <p>We design a general framework for print quality detection and analysis of customer content. This print quality analysis framework inputs the original digital image saved on the computer and then the scanned image. This framework includes two main modules: image pre-processing, print defects feature vector extraction, and classification. The first module, image pre-processing, includes image registration, color calibration, and region of interest (ROI) extraction. The ROI extraction part is designed to extract four different kinds of ROI from the digital master image. Because different ROIs include different print defects, for example, the symbol ROI includes the text fading defect, and the raster ROI includes the color fading defect. The second module includes different ROI print defects detection and analysis algorithms. We classify different ROI print defects using their feature vector based on their severity. This module proposed four important defects detection methods: uniform color area streak detection, symbol ROI color text fading detection, raster ROI color fading detection using a novel unsupervised clustering method, and raster ROI streak detection. We will introduce the details of these algorithms in this thesis. </p> <p><br></p> <p>We will also show two other interesting print quality projects: print margin skew detection and print velocity simulation and estimation. Print margin skew detection proposes an algorithm that uses the Hough Lines Detection algorithm to detect printing margin and skew errors based on factual scanned image verification. In the print velocity simulation and estimation project, we propose a print velocity simulation tool, design a specific print velocity test page, and design a print velocity estimation algorithm using the dynamic time warping algorithm. </p>