Understanding HCI Requirements: Expertise and Assistance to Imagery Analysis

Understanding HCI Requirements: Expertise and Assistance to Imagery Analysis I.S. MacLeod, A.J. McClumpha and E. Koritsas 1. Introduction Reconnaissance, Surveillance and Target Acquisition (RSTA) systems are increasingly based on advanced technology and are required to handle a diversity of sensors to exploit the associated forms of collected imagery. The future real time collection and dissemination of imagery will require co-operative use of multi-user workstations that have the capacity to display a variety of different forms of imagery (for example, Synthetic Aperture Radar (SAR), Moving Target Indicator (MTI) Radar, Electro Optic (EO) Sensors, and Infra-Red (IR)). To illustrate the variety of imagery that the Image Analyst (IA) has to analyse, Figures 1, 2 and 3 are provided in illustration. Each figure has some explanatory text included with the title. Figure One - An MTI Scene. (Thick lines and dots are MTI returns; thin lines depict roads and coastlines. It can be readily appreciated that improved map cultural detail, contrasted against MTI returns, would assist analysis of this image). Currently, technology offers important assistance for the collection, display, storage, display and dissemination of imagery. However, only minimal technical assistance is given to the IA who is required to analyse and exploit the presented imagery. It is clear that in current 'technologically sophisticated systems' there is only limited support to the skilled performance of the cognitive and perceptual analyses that are required to interpret imagery. Furthermore, the introduction of this new technology has changed the nature and methods of imagery analysis and the associated work practices that the IA needs to undertake. These new work practices are also being relocated into new work domains where the maintenance of situational awareness is becoming more important (Reference 1). However, few studies have addressed the problem of taking a user-centred perspective to both qualitatively and quantitatively explore the cognitive and perceptual aspects of imagery analysis. The work reported in this paper is part of an ongoing Human Factors (HF) programme to better understand the cognitive and perceptual aspects of the imagery exploitation tasks so that guidance can be offered to make efficient use of technology and to enhance the productivity of the analyst. In recent years the UK Centre of Human Sciences (DERA CHS) has conducted several research concerning the use of VDU and computer technology to assist the process of image analysis (References 2 & 3). These studies attempt to identify and scope the expertise involved in the human cognitive and perceptual tasks that are crucial to the efficient and effective analysis of images. The objective of this paper is to present and discuss a few of the findings from studies and to indicate to the better understanding of aspects of human performance in the analysis of imagery. This improved understanding will help provide a more user-centred basis for the future determination of HCI standards and guidelines to facilitate performance in image analysis. 2. Guideline Requirements Essentially, guidelines should give advice on what information is needed, its presentation, form, timeliness, and the rate of its display. Many HCI related guidelines have been produced in recent years, their content depending on the subject area and the agency commissioning the guideline work. However, no HCI guidelines have been produced that are focused on the display and analysis of imagery. Figure 2 - SAR Example - Sea Empress Oil Spill near Milford Haven, UK. (The course 30 metre resolution of this satellite image still allows meaningful depiction of the spill size). One influential book produced on the problems of image analysis and target detection, and still pertinent to this day, covered the problems of image analysis from the perspective of human perception and the many diverse influences on that perception (Reference 4). However, this book failed to attempt any delineation of the cognitive contribution to analysis. It follows that prior to the production of any HCI guidelines related to imagery analysis, the fundamentals of the associated imagery analyst's cognitive based expertise must be discovered, and pertinent analysis tasks understood. This work is essential to produce the source material necessary for the production of guidelines or for the design of HCI based tools to assist the IA's analysis tasks. Part of the problem is the determination of the form of skill requirements and strategies most useful to the performance of IA tasks. These skill requirements should be considered with regard to personnel selection, training, and the appropriateness or level of assistance given to the IA through system and HCI design. In the UK, the selection process for the IA is primarily based on vision capabilities (stereoscopic vision is deemed to be essential) and aspects of applied intelligence. UK based ab initio training of IAs emphasises a procedural approach to image analysis; but in contrast, recent HF studies have shown that an important manifestation of IA expertise is an almost automatic approach to analysis tasks. Another manifestation of expertise is the variety of approaches used to the analysis of imaged targets. These findings are all of importance to post ab initio continuation training. Therefore, as well as studying the nature of the expertise required of the future IA (for some discussion on the nature of skill and expertise see References 5 & 6), it is necessary to study the work factors that influence the application of that expertise. These are the factors that should be the primary influence on HCI design. Examination of this topic requires the use of a variety of approaches and techniques. Domain experts have participated in a number of studies to explore the performance based aspects of the cognitive and perceptual interpretation of imagery. These studies have been conducted using equipment currently operational within the UK Royal Air Force. 3. Methods and Implications The recent approaches to the problem have covered the following areas: classification of IA skill; classification of targets; understanding of the implications to image analysis of imagery types; classification of IA tasks. These approaches have been supported through a combined use of many methods, mainly qualitative in nature. They have included the use of such as concurrent and retrospective verbal protocol analyses; repertory grid techniques; subjective workload assessments; method of limits; concept mapping; cognitive probes; individual and group debriefs. The studies have indicated that there are important differences in the way that domain experts with differing levels of expertise carry out the task of target interpretation. For example, experts do not apply similar analysis procedures to each imaged target but individually vary the procedure depending upon the particular characteristics of the target, its familiarity, and the form of the imagery. This finding is important because it demonstrates that there is not likely to be a simple explanation of how to provide technological support to the domain expert for imagery exploitation. The finding also emphasises the need to establish a cognitive based assessment of the nature of image analysis tasks. Furthermore, it is also clear that the form of required system support to the IA may need to be qualitatively and quantitatively different between IAs with different levels of experience. In addition, there is an important difference between the usage of an image based HCI and an alpha-numeric HCI. With image based HCI the input is the image, the work is primarily conducted on the displayed image, but the output is normally alpha numeric. With alpha numeric based HCI the input and outputs are normally based on alpha-numerics, and possibly associated graphics, and the work is normally only associated with the displayed information. It has been found that the greater the experience of the IA the more they require all image manipulation controls to be through mouse / roll ball and cursor so as to allow continuity of attention and the analysis through an image display occupying the whole of the display area. Novices and the less skilled IAs initially prefer a set piece screen control panel to occupy part of the display in support of their more procedural approach to image analysis. This preference rapidly diminishes with experience. All IAs prefer the use of icons as a method over menus to select ancillary work applications. Figure 3 - IR Image of an Oil Refinery (Various levels of liquid in the tanks is discernible as is the increasing distortion of the image towards the horizon). 4. Conclusion The skills and developed expertise of the IA are largely cognitive and have yet to be fully understood. Modern technology assists in the display and development of the imagery but gives little assistance to the actual analyses of images by the IA. Implications for the development of suitable HCI for the IA are that the displayed image is considered to be sacrosanct and that the image manifestation must make optimum use of the available display area. Therefore, the system/image control facilities should accommodate as little of the display space as possible and that any additional analysis facilities should be accessed as appropriate through simple Icon selection. However, the less expert IA still needs dedicated screen and image manipulation controls as a steppingstone towards their expertise with the use of the analysis system and the exploitation of various forms of imagery through that system. Future guidelines for imagery HCI should approach these points. References MacLeod, I.S., Taylor, R. M & Davies, C.D. (1995) Perspectives on the Appreciation of Team Situational Awareness, in Proceedings of the International Conference on Experimental Analysis and Measurement of Situation Awareness, Embry-Riddle Aeronautical University Press, Florida. McClumpha, A.J. & MacLeod, I.S., (1996), Skills and Understanding in Imagery Exploitation, 1996 SAR/SLAR Steering Committee Meeting, Berlin, Germany, 17-19th October 1996. MacLeod, I.S. (1996), Final Report: Analysis of IAs Cognitive and Perceptual Activities, undertaken under UK DERA CHS Contract No CHS3/5038. (AeroSystems International 0543S/1/TR.1-1 dated June 1996). Jones, D.B., Freitag, M., & Coolyer, S.C., (1974), Air to Ground Target Acquisition (TA) Source Book: Review of Literature, Office of Naval Research, NR 196/121. Boy, G (1995) Knowledge Elicitation for the Design of Software Agents, in Handbook of Human Computer Interaction, 2nd edition, Helander, M. & Landauer, T. (Eds), Elsevier Science Pub., North Holland. 6. Ericsson, K.A. & Staszewski, J.J. (1989), Skilled memory and Expertise: Mechanisms of Exceptional Performance, in Complex Information Processing: The Impact of Herbert A. Simon, Klahr, D. & Kotovsky, K. (Eds), Lawrence Erlbaum, NJ. esign of Computing Systems Social and Ergonomic Considerations (Smith, M. J., Salvendy, G. & Koubek, R.J. (Eds), (1997), Elsevier. ISBN: 0 444 82183 X Paper pps: 559-562