Development and Research Skills: A Dissertation on Satellite image Classification - Essay Example

Number Development and Research Skills: A Dissertation on Satellite image ification. Submitted in partial fulfillment for the M.Sc. in Computing (2012) Workshop Tutor. Table of Contents Table of Contents 1 AIMS AND OBJECTIVES 2 REVIEW OF CURRENT STATE OF PROPOSED AREA 4 MAJOR MILESTONE DELIVERABLES 6 8 SCIENTIFIC RISK ANALYSIS 8 RESOURCES NEEDED AND ARCHITECTURE 9 ETHICAL, LEGAL, PROFESSIONAL ISSUES AND ACADEMIC MISCONDUCT 11 REFERENCE LIST 14 AIMS AND OBJECTIVES The study of image classification is a crucial undertaking as it assigns data items into groups or classes that elaborate certain scientific concepts like spectral information that is inherent in the image (Boyer and Umsallam 2011). The use of satellites to observe items have created a clear perspective in the research field with improvement of imagery and observation without coming into contact with the object under consideration (Espinola et al 2008). This research study has several aims that try to ascertain and investigate the relevance and progress of image classification through the use of algorithmic concept of remote sensing. Through the understanding of ground calibration targets, the correction of satellite sensor data with regard to optical regions is enhanced (Mountrakis 2008). Through the processes of selecting suitable targets, subjective and time proves to be a task. Therefore, as a researcher, improving the concept of image classification is highly critical as it perfects the approaches and techniques of image processing; the understanding of the probable working ways of satellite image classification is a crucial undertaking. Looking at the specific aims and their objectives regarding the research study, the followed can be deduced. The first aim regards the understanding of the concept behind image classification, which is critical, with regard to this research study as the researcher was able to ascertain the background information regarding the developments made, as well as comprehends them. This critically initiates a research interest in trying to ascertain the objectivity behind the concept of using satellite image observation. Satellite image classification is critical in the sense that it helps bring to focus scientific observations that could not have been observed by the researcher (Mountrakis et al. 2009). The images under focus are presented without necessarily coming into contact with the object. Having understood the concept, the researcher gets the contextual nature and basis to devise and recommend for crucial improvements within the field of satellite image classification. The second aim of this research study incorporates the review of issues hindering the progress to achieving success satellite image classification even with the most distant objects. A clear example is the observation of terrestrial bodies and natural satellites. Failure to obtain clear images leads to poor interpretation of the data imagery. According to Reusch (1999), firstly, analysis of the data influence regarding representation and cluster determination methods are crucial in classification accuracy. Therefore, the mode of data collection significantly contributes towards the kind of image that is to be classified (Benediktsson et al. 1990). With a clear understanding of the hindrances to effective data classification, researchers can amount to coming up with effective modes of image classification like improved satellite devices to enhance imagery (Weber 2006). Lastly, future advances of image classification are intensely critical in the research field; therefore, this research study aims at ascertaining the probable advances that are imperative with current global advancement. Flowchart representation of the specific aims and their objectives regarding the aims to be met REVIEW OF CURRENT STATE OF PROPOSED AREA Image classification has made considerable progress with regard to the techniques and approaches applied. Designs of automated systems dedicated to satellite image classification have been given high levels of concern (Reusch 1999Espinola et al. 2008). With crucial results being achieved, numerous classifiers founded on the spectral analysis of entity pixels, have been proposed. With all the progress, the limitations encountered while using spectral information are apparent, and mostly the classification results are unsatisfactory (Richards 1982). The assessment and monitoring of the state of the earth’s surface is critical to global change research (Jung et al. 2006; Lambin et al 2001). For example, the utilisation of classifying and mapping of vegetation has proved to be a pivotal technical task in the management of natural resources (Richards 1995). To achieve high accuracy, integration of extra data has been evaluated. The multisource classification model has been based on pixel-by-pixel classification techniques combining spectral information together with diverse varieties of data linked to individual pixels like multi-temporal data and others (Richards et al. 1982; Russo and Ramponi 1994; Clark et al. 1996). With the improvement of satellite and remote sensing modus operandi, considerable data from airborne/satellite sensors have been made available (Boyer and Unsallam 2011). For example, multi-sensor image fusion seeks to combine information from different images so as to come up with more inference than can be plagiaristic from a single sensor. Few surveys have been made have been published with regard to overviews of the progress and the current state of the art of image classification (Espinola et al. 2008). On the same note, elaborate discussions on the recent data developments of multi-sensor data fusion in remote sensing fields have not been availed. Therefore, there is high urge to incorporate the new measures in trying to expound on new developments in the field of image classification, which can only be achieved via extensive and elaborate research studies. An overview categorisation of the fusion algorithms, MAJOR MILESTONE DELIVERABLES The dissertation report will entail various stages from the preparation to the final submission of the dissertation. This section aims at laying down the tasks behind a successful research undertaking and final submission of the research study. The information have been summarised by use of a Gantt chart entailing all the key tasks from week one to week thirteen (Rugg and Petre 2007). The research work commences with the selection of an area of study and then a topic for research drafted. This topic had to be presented in class and agreed upon with the tutor prior to preliminary investigations. This was followed by the presentation of an A4 slide outlining the scope of topic under consideration together with a project plan detailing steps to be undertaken in the production of the report. At the fourth week, enough preliminary research is through, and a proposal of approximately 500 words is ready. This will outline the projects undertaking and background information, as well as entail the scope of the study. After the proposal presentation, a report into the dissertation is outlined. This identifies the problem under study, as well as the topic, and the intended solution. On the same note, literature on the topic is discussed together with a review of an academic paper on satellite image classification. The review under consideration outlines the development of classifiers with regard to various techniques. After a thorough review of a key paper, personal recommendations regarding future research work are imperative. Lastly, a personal development plan will terminate the report for submission at week 9 together with a poster for the research study. Finally, the dissertation or the project report (coursework two) will be prepared for submission to the graduate office at week 13. This will incorporate all the conceptual and research work undertaken for the study under consideration. This is highly crucial in ascertaining the findings and the recommendations of study at hand with respect to satellite image classification. SCIENTIFIC RISK ANALYSIS The scientific risk analysis of the research project falls in the use of electromagnetic spectra that can be associated with environmental hazards. Though the project can be called innovative where no comparable systems exist, project can be called innovative when no comparable system exists, it can be termed risky with regard to the researcher not having a contrite surety of the eventual outcome. Various projects carried out by some organisation were innovative but at the same time extremely risky. Further, the interference of individuals’ privacy by the emissions emanating from the observatory nature of the instruments used, pose a challenge to the communities, and also to the researcher in the sense of not attaining the set goals. Other risks emanating with this scientific study is the technological hitches that may render the instruments and programs insufficient. This regards to the data obtains thereby, demeaning or lowering the quality of the results, if the finding pertains the lives of humans, this may lead to anxiety whenever there is irrelevant release of information or data. The risk avoidance strategy will be through the adherence to the set regulations and ethical standards. This will safeguard the data collection and imagery presentation. The factual concept regarding the risk control would be through the human collections and creativity to alleviate the same. RESOURCES NEEDED AND ARCHITECTURE The recent growth in technological advancement has seen the incorporation of computer science technology in the image processing to influence and enhance the imagery of data. Softwares like ESRI has a leading GIS software provider globally notably the ArcGIS product (Russo and Ramponi 1994). Erdas is used in remote sensing and photogrammetry suite called LPS. Other packages like PCI or ENVI could be used, though more evaluation is critical. The softwares are critical in the sense that they conduct data acquisition and analysing ad further classify them to obtain images (Weber 2006). Other requirements are the directions provided by human operators so as to perform an image interpretation regarding to various conditions. The conditions require the definition of the operator. This is orchestrated by the fact that digital image classification of various material on the earth’s globe have different spectral characteristics (Russo and Ramponi 1994).. The softwares are crucial in separating and grouping of the spectral qualities and at the same time identify the critical aspect of imagery and allocation of classes to the pixels (). The essence for the softwares and hardwares in this research are in the determination of the spectral signature of the pixels (Jung et al. 2006). The safety and confidentiality of the assignment falls in the fact that the sensitivity of the factual data ought to be safeguarded by the researchers. This is done by ensuring that the project is recognised by the relevant authorities. A representation of the image to inform the remote sensing software. http://rst.gsfc.nasa.govt/Front/tofc.html Hardware in simulation strategy http://www.gepards.at/projects.htm ETHICAL, LEGAL, PROFESSIONAL ISSUES AND ACADEMIC MISCONDUCT The research study under consideration is truly involving and sensitive. There are principled, official, and expertise issues that need much consideration prior and in the course of the research work. On the same note, the academc misconduct that have to be evaded will be highlighted have to be discussed in this section. The satellite imaging industry falls under the jurisdiction of the U.S. Commerce Departmet and the responsibility of regulation and licensing falls with NOAA (Jung et al. 2006). The researchers conducting studies in this area must adhere in writing to the obligation set and violations lead to revocation of licence, retention in the future or even civil penalties. Regarding the ethical issues, this research has ensured that adherence to set ethical standards with respect to electromagnetic properties, researches is enhanced. The use of satellite imagery should never compromise the national security or lead to the exposure of sensitive operations (Lambin et al. 2001). Ethical issues in this regard are connected to the controlling any arising suspicious behaviours that may lead to legal tussles involving the satellite research bodies and the authorities. With this regard, the GPS satellite tracking come handy in assisting to identify those violating the set standards (Reusch 1999). Further, data management issues within the study team is very critical, and pertains to the subjects having the legal rights to the data and who retains the information analysed. Academic conduct entails the upholding of decency and this is evidenced with the search for perfect data. Fabrication and falsification of any research data results is a serious form of misconduct. The above should not deceive through data alteration, circumvention of existing laws and the transgression of reasonable and legitimate expectation of privacy. Misinformation and manipulation have been problems serious problems regarding the imagery; whether the image or information came from outer space or form unknown source, as a professional, it is particularly crucial to always confirm the information before letting it out to the public (Clark et al. 1996). This maintains and preserves the integrity of research and its finding in the future. With regard to geo-spatial data, particular efforts should be made to safeguard its privacy as the information can be use to undermine interests of humans and thereby, the agreements made to keep the data from a public domain should be upheld (Boyer and Unsallam 2011). The following give clear examples of research approaches in geospatial technologies that could raise privacy and issues to do with confidentiality, thus ought to be undertaken with considerable care. First is the automated tracking of locations and movements of vehicles and individuals. Secondly is the use of images from satellites, aircraft or even ground sensors which possess sufficient resolution to identify individuals or vehicles and lastly the use of geographic locations, regarding form of coordinates or street addresses used to link diverse sources of data information pertaining personal nature (Boyer and Unsallam 2011). This calls for vigilance among researchers and workers exploiting the concept of satellite image classification concepts and continuance in their education on how to report their findings and the sources used to do the reporting. This will ensure that satellite imaging remain useful to the world of research, also to areas of professionalism utilising the concept like the mass media. REFERENCE LIST Benediktsson, J.A., Swain, P.H. and Ersoy, O.K., 1990. Neural Network Approaches versus Statistical Methods in Classification of Multisource, Remote-Sensing Data: IEEE Trans Geosci. Remote Sensing, Vol. 28(4), pp. 540-552. Boyer, K. L. and Unsallam, C., 2011. Multispectral Satelite Image Understanding: From Land Classification to Building and Load Detection. Springer-Verlag: London 2011. Clark, P., Feng, C., Martin, S. and Fung, K., 1996. “Improving Image Classification by Combining Statistical, Casebased and Model-based Prediction Methods,” Fundamenta Informatica, Vol. 30(3-4) pp. 227-240. Espinola, M., Ayala, R., Leguizamon, S. and Menenti, M., 2008. Classification of Satellite Image s Using the Cellular Automata Approach: Communications in Computer and Information Science. Vol. 19, pp. 521-526. Jung, M., Churkina, G. Henkel, K, et al., 2006. Exploiting Synergies of Global Land Cover Products for Carbon Cycle Modelling. Remote Sens Environ. Vol. 103, pp. 534-553. Lambin, E. F., Turner, B.L. and Helmut et al., 2001. The causes of Land-use and land cover change: Moving Beyond the Myts. Glob Environ Chang, Vol. 11, pp. 261-269. Mountrakis, G., Watts, R., Luo, L., Wang, J., 2009. Developing Collaborative Classifiers using an Expert-based Model. Photogrammetric Engineering and Remote Sensing, 75(7):831-844. Mountrakis, G., 2008. Next Generation Classifiers: Focusing on integration frameworks. Highlight article for Photogrammetric Engineering and Remote Sensing, 74(10):1178-1180. Reusch, B., 1999. “Extended Methods for Classification of Remotely-Sensed - Images, Based on ARTMAP Neural-Networks,” in Computational Intelligence: Theory and Applications: International Conference, 6th Fuzzy Days Dortmund, May 1999 Proceedings. New York: Springer. pp. 206-217. Richards, J., 1995. Remote Sensing Digital Image. Berlin: Springer- Verlag, pp. 265- 290. Richards, J., Landgrebe, D. and Swain, P., 1982. “A Means for Utilising Ancillary Information in Multispectral Classification, “Remote Sensing of Environment, Vol. 12, pp. 463-477. Rugg, G. andPetre, M., 2007. A Gentle Guide to Research Methods, Open Berkshire: University Press, McGraw-Hill. Russo, F. and Ramponi, G., 1994. “Fuzzy Methods for Multi-sensor Data Fusion, “IEEE Trans Instrumentation and Measurement, Vol. 43(2) pp. 288-294. Weber, K., 2006. Challenges of Integrating Geospatial Technologies into Rangeland Research and Management. Rangeland Ecol Manage. Sharp, D. Peters, J. and Howar, K., 2002. The Management of a Student Research Project. Burlington: Gower Publishing Company. Read More