As early as the 1970’s it was determined that gait, or the “manner of walking” 1 is an identifying feature of a human being. 2 Since then, extensive research has been done in the field of computer vision to determine how accurately a subject can be identified by gait characteristics. This has necessarily led to the study of how various data collection conditions, such as terrain type, varying camera angles, or a carried briefcase, may affect the identifying features of gait. However, little or no research has been done to question whether such conditions may be inferred from gait analysis. For example, is it possible to determine characteristics of the walking surface simply by looking at statistics derived from the subject’s gait? The question to be addressed is whether significant concealed weight distributed on the subject’s torso can be discovered through analysis of his gait. Individual trends in subjects in response to increasing concealed weight will be explored, with the objective of finding universal trends that would have obvious security purposes. 1. PREVIOUS RESEARCH Characteristics of human gait have been studied extensively for the purposes of subject identification. The U.S. Army Research Laboratory has also extensively studied the physiological, biomechanical, and medical aspects of load carriage using packs. Research involving the effects of adolescents carrying backpacks has also been performed. However, experiments and data analysis designed specifically for the detection of load carriage are non-existent. Still, trends discovered in previous, related research suggest consistent characteristics that may be used in the visual detection of load carriage. The Army's research indicates the duration of the swing phase (foot in the air) decreases with increased load. This also results in an increase in the time of double support with load. Forward inclination of the body trunk, as well as knee flexion, also increase significantly with load. 3 However, such trends would require a comparison of the subject walking unencumbered with the subject walking with a load to identify when the load was present. It remains yet to be seen if a subject can be identified as carrying a significant load without a comparison to their unencumbered gait characteristics. Such detection requires the use of metrics that can be standardized across subjects. It has been shown that the subject's apparent height (the vertical distance between the top of the head and the ground) is a very consistent factor in a given set of circumstances. 4 This suggests that some measurements that increase linearly with subject height (stride length, maximum hand separation, etc.) may be standardized as ratios with apparent height. Various methods of visual analysis have been used to extract gait characteristics. It has been shown that silhouettes (obtained by background substitution) provide enough periodic information for fairly consistent identification. 5 Conventional markers attached to key joints have also been used extensively, although this method obviously does not work with an unknowing subject. A more versatile approach to visual analysis has been to dynamically identify body parts based upon a computer model. 6 The gait characteristics are then extracted from these estimated models. This technique has the advantage of not requiring markers or any additional equipment. 2. EXPERIMENT DESIGN This study used the method of visually determining gait characteristics with a set of markers attached to key locations on each subject. While this is not as versatile a method of visual analysis as some methods previously mentioned, our goal is not to develop a robust system for the identification of an uncooperative subject carrying a load. Rather, the goal is to address the question of whether it is even possible to determine load carriage through visual analysis of gait characteristics. For this reason, the use of markers, which provide a simpler, more consistent, and more thorough method of analysis were used. Each subject had a total of 11 markers attached to them on the following locations: top of head, near shoulder, near elbow, near wrist, near hip, near knee, near ankle, near heel, near toe, far heel, and far toe ('near' and 'far' being relative to the camera). These locations were chosen to provide key gait characteristics such as stride, cadence, and joint angles, among others. A straight walkway with two cameras set up perpendicular to it a great distance away was used. This allowed simple two-dimensional analysis of the subjects that could be possible from something like a security camera. The cameras were placed half the length of the walkway apart, each capturing one side of it. The cameras were allowed to rotate freely about their vertical axes and were zoomed in so they just comfortably fit the entire subject within their field of view. Both cameras start with an orientation towards the left half of their side of the walkway and track the subject as he moves through their respective halves of the walkway. While a changing viewing angle complicates calculations, a large distance between the cameras and the walkway ensures the angle from perpendicular will always be small (in our particular setup, it was never more than 6 degrees). Therefore, the cameras are treated as being constantly perpendicular to the subject’s direction of travel for the purpose of analysis. To simplify the problem being addressed, the study was designed to use as simple an environment as possible. This meant an indoor location with a firm, even walking surface and consistent lighting. Also, all test subjects were male because this experiment does not explore the differences in gait characteristics in males and females. In addition, each subject completed at least 5 full gait cycles in each sample to ensure the capture of a steady state sample.