QUANTIFYING HAIR MOTION 373 monochrome video camera that captures 70 frames per second with a resolution of 1,850 × 2,300 pixels and 8-bit digitalization. The videos are saved in AVI format, without any compression to retain all image information for subsequent processing. Modifi ed software allows for the control of experimental setup, visualization of captured videos, and a variety of post-experiment imaging analyses. For example, experiments can be setup at a single specifi c frequency or automatically run across a range of values, and tress motion can be replayed in real time or slow motion. At this early stage of the work, we have considered a variety of analyses to gather informa- tion on tress movement. Figure 2 shows the result of tress edge detection that allows for a skeleton of the swatch to be determined for each video frame. In this way, it is possible to derive measures for the volume and shape of the tress bulk during motion. Moreover, the homogeneity of the tresses can be assessed during motion by looking at relative levels of wispy fl yaway fi bers—as per the purpose of the initial device. RESULTS Initial experiments were performed using clean, straight tresses of healthy Caucasian hair. Tresses contained approximately 3 g of hair and measured 20 cm in length by about 1 cm in width. These tresses gave rise to a natural oscillating motion in their clean, untreated state but the negative impact of fi xative polymers (styling products) and relatively heavy coatings (oils) was visually evident in early “feel-out” experiments. The natural swinging motion of the hair tips was considerably dampened after oil treatment and essentially eliminated by the styling treatment. Thus, the amplitude of the swaying associated with the tress tips appeared an instinctive measure of this motion. Figure 2. Skeletal image of moving hair tress allowing evaluation of displacement for ends relative to anchor point.

JOURNAL OF COSMETIC SCIENCE 374 Figure 3 shows results for a straight Caucasian hair tress oscillating at a frequency of 1 Hz. The oscillating tress displaces from the center point by approximately ±20°—resulting in a total amplitude of motion of about 40°. For comparison purposes, an identical experi- ment using decidedly curly/frizzy hair resulted in an amplitude of 28°, whereas Cauca- sian hair treated with a hair spray essentially gave a value of zero. Thus, a direct correlation between the magnitude of the measured amplitudes and the perceived motions of these tresses might be presumed—however, considerably greater complexity exists. The considerable infl uence of the oscillating frequency is another observation that arises from watching hair with natural and unnatural motion. In clean tresses, the nature of the motion varies signifi cantly with changing frequency however, this does not happen in hair treated with a hair spray. As possibly anticipated, application of progressively higher fre- quencies of motion initially produces higher amplitudes in clean tresses—but only up to a point. Eventually, application of still higher frequencies ultimately produces a decrease in this parameter—presumably as hair motion in one direction is stifl ed by the tress now being pulled in the opposite direction, before motion has fi nished in the original direction. Figure 4 shows examples of amplitude versus frequency curves for the aforementioned Caucasian, frizzy, and hair spray–treated hair. As already noted, the independence of the hair spray–treated hair on frequency is considered an aspect of unnatural motion. This graph illustrates how conclusions based on a single frequency can be misleading. That is, experi- ments at 1 Hz showed straight Caucasian hair producing a higher amplitude of motion than frizzy hair however, this trend is seen to reverse at higher frequencies. A value for the maximum attainable amplitude (Amax) for any particular test sample and the fre- quency at which it occurs (fmax), potentially become useful characterization parameters. HAIR TRESS EFFECTS In most in vitro hair testing experiments, the nature of the results can be greatly infl uenced by the properties of the hair itself and the size/shape of tresses. For example, when per- forming extremely popular instrumental combing experiments, it is desirable to initially Figure 3. Motion amplitude for a Caucasian hair tress under the application of a 1-Hz oscillating frequency.