Data Analytics, Computational Statistics, and Operations Research for Engineers

Methodologies and Applications

Debabrata Samanta, SK Hafizul Islam, Naveen Chilamkurti, Mohammad Hammoudeh, Debabrata Samanta, SK Hafizul Islam, Naveen Chilamkurti, Mohammad Hammoudeh

• Publisher: CRC Press
• Year: 2022
• ISBN:
  9781000550467
• Edition: 1
• Topic: Ciencia de la computación
• Subtopic: Minería de datos

With the rapidly advancing fields of Data Analytics and Computational Statistics, it's important to keep up with current trends, methodologies, and applications. This book investigates the role of data mining in computational statistics for machine learning. It offers applications that can be used in various domains and examines the role of transformation functions in optimizing problem statements.

Data Analytics, Computational Statistics, and Operations Research for Engineers: Methodologies and Applications presents applications of computationally intensive methods, inference techniques, and survival analysis models. It discusses how data mining extracts information and how machine learning improves the computational model based on the new information.

Those interested in this reference work will include students, professionals, and researchers working in the areas of data mining, computational statistics, operations research, and machine learning.

1 Hyperspectral Imagery Applications for Precision Agriculture A Systemic Survey

Chanki Pandey, Yogesh Kumar Sahu, Prabira Kumar Sethy, and Santi Kumari Behera

DOI: 10.1201/9781003152392-1

CONTENTS

1.1 Introduction

1.1.1 The Main Contribution of the Chapter

1.2 Hyperspectral Imaging Technology

1.3 Agricultural Applications

1.3.1 Soil Analysis

1.3.2 Contaminants and Nutrient Estimation

1.3.3 Inland Water and Moisture Estimation

1.3.4 Crop Yield Estimation

1.3.5 Plant Disease Monitoring, Insect Pesticide Monitoring, and Invasive Plant Species

1.3.6 Agricultural Crop Classification

1.4 Conclusion and Future Scope

References

1.1 INTRODUCTION

Due to the current world’s population, the demand for food will continue to rise. Unfortunately, environmental waste has reduced the percentage of arable land. As a result, farming and livestock yield activities are gaining prominence to satisfy the overflowing food demand. Precision agriculture is a method for optimizing productivity to fulfill increasing food requirements while minimizing economic and environmental costs related to food production. Precision farming includes extensive knowledge of temporal and spatial changes in crop conditions collected by remote sensing (Gevaert et al. 2015). The development of technologically advanced instruments for agricultural applications is important in helping farmers make crop health and resource management decisions. In addition to helping farmers efficiently use herbicides and pesticides, successful hyperspectral imaging in precision agriculture also offers insights into the current phase of crop growth and health. Precision agriculture uses cutting-edge technology to make farming more regulated and precise, and imaging technologies, especially multispectral and hyperspectral imaging, are gaining prominence worldwide. According to the most recent Persistence Market Research (Persistence Market Research 2016) report, the demand for global imagery technologies in the precise agriculture market is expected to grow at a 9.0% CAGR from 2016 to 2024 at a market revenue of $1,165.9 million. According to the researchers, the precise agriculture imaging technology market produced $567.4 million in 2016, a 6.2% growth over 2015.

In large application areas such as mineral identification, environmental analysis, precise agriculture, and urban planning, the hyperspectral images can differentiate between different artifacts and physicals. Among other end-use sectors, forestry and agriculture are expected to have the largest market share in the hyperspectral imaging markets (Fact.MR 2019), as shown in Figure 1.1. In forestry and agriculture, hyperspectral imaging is used for various applications such as plant recognition, seed output analysis, weed mapping, and forest management. Furthermore, the amount of data gathered on farms from sensors has risen significantly over the last decade. Hyperspectral service providers have synchronized offers for data optimization and applications for farmers.

FIGURE 1.1 Usages of hyperspectral imaging (HSI) by the end-use industries.

Nutrient crop surveillance, water stress, disease, insect infarction, and overall plant health are key components of effective farming operations. “The assurance of appropriate spatial and spectral knowledge for the non-destruction assessment of food and agricultural products is limited by traditional optical sensing technologies such as imaging or spectroscopy” (X. Li et al. 2018). In general, traditional imagery cannot obtain spectral information, and measurement by spectroscopy cannot cover vast areas of the sample. The commonly used fruit and vegetable sorting vision systems are based on a color video camera that emulates the vision of the human eye by taking photographs with red, green, and blue (RGB) wavelengths (Costa et al. 2011; Cubero et al. 2011). As a result, they are restricted to analyzing scenes. They are normally unable to detail the exterior or internal structure of the materials or spot flaws or changes whose color is close to the color of the tough skin. Furthermore, conventional methods of tracking fruits and vegetables requiring analytical techniques are too time-consuming and costly and necessitate sample destruction.

Over the last few decades, optical sensor systems have evolved as scientific instruments for precision agriculture, thanks to the exponential advancement of information science and the analysis of images and patterns. In particular, spectroscopy and imaging techniques have been extensively researched and developed by incorporating a method that can obtain a spectral variance spatial map, which results in many popular applications in agriculture. The advancement of aerial and ground-based hyperspectral and multispectral imaging technology has been a significant step in the extension and practical implementation of precision agriculture techniques. In addition to its predictive capabilities, this technology has enabled evaluating crop stresses, characterization of soils and vegetative cover, bruise detection in fruits and vegetables (Z. Du et al. 2020), moisture content estimation (Z. Wang et al. 2020), and yield estimation (Y. R. Chen et al. 2002). Some of the advantages of hyperspectral and multispectral imaging are low costs (compared with conventional scouting); reliable performance; easy to use, quick, nondestructive, and highly precise evaluations; and applications of wide variety. A conventional spectral image comprises a series of monochromatic pictures that correspond to different wavelengths. Hyperspectral image systems have a natural benefit over conventional machine vision (Y. R. Chen et al. 2002) and even human vision. Any appearance characteristics difficult or hard to extract with conventional computer vision systems can be extracted using hyperspectral imaging systems. The quality and safety evaluation of various agriculture and food products has proved to be a vast application for hyperspectral imaging, for example, lamb (Fowler et al. 2015; Kamruzzaman et al. 2011; Pu et al. 2014), fruit (S. Chen et al. 2015; Nogales-Bueno, Rodríguez-Pulido et al. 2015), fish (Kimiya et al. 2013; Menesatti et al. 2010; Wu and Sun 2013; H. Zhang et al. 2020), cereals (Bianchini et al. 2021; Caporaso et al. 2018; Feng et al. 2019; Fox and Manley 2014; Manley et al. 2009; Paliwal et al. 2018; Qiu et al. 2018; Rabanera et al. 2021; Rathna Priya and Manickavasagan 2021; Sendin et al. 2019, 2018; Wakholi et al. 2018; Zeng et al. 2019), milk (Forchetti and Poppi 2017; Jawaid et al. 2013; Lim et al. 2016), pork (D. Barbin et al. 2012a, 2012b, 2013; Jiang et al. 2021; Silva et al. 2020; Xie et al. 2015), beef (Dixit et al. 2020; ElMasry et al. 2013; Naganathan et al. 2008), carcass of poultry (Falkovskaya and Gowen 2020; Nakariyakul and Casasent 2009; Nyalala et al. 2021; Park et al. 2007), veggies (Gowen et al. 2008; Ji et al. 2019; Nguyen Do Trong et al. 2011; Nogales-Bueno, Baca-Bocanegra et al. 2015), and so on.

1.1.1 THE MAIN CONTRIBUTION OF THE CHAPTER

1. The basic concepts of hyperspectral imaging technologies are covered.
2. The basics of hyperspectral imaging for food safety inspection are discussed.
3. The benefits and drawbacks of hyperspectral imaging technology are explored.
4. A review of hyperspectral image- processing techniques is given.
5. The current chapter highlights and outlines many important topics in key research concerning hyperspectral agricultural imagery applications.

In this brief study, we examined the past of hyperspectral imaging, imaging systems, precision agriculture applications, and hyperspectral data-processing techniques. To collect precise agriculture data, researchers can use hyperspectral imaging systems. Given the basic literature understanding, the topics of hyperspectral image processing are briefly explored in agriculture. The review covers both fundamental analyses of research and somewhere within the social facets of research. This analysis mainly seeks to identify potential works addressing the key issues and speed up deployable end-user solutions to achieve existing global goals for sustainable development. Section 2 contains simple hyperspectral image information and tools for interpretation. Section 3 analyzes the benefits and disadvantages of the various approaches used for hyperspectral imaging in precision agriculture. Section 4 provides a concise summary and concludes this chapter with the future scope of the hyperspectral imagery system (HSI) for precision agriculture.

...