Abstract: A device for finding the centroid coordinates of a figure that can calculate the centroid from a single time measurement of the figure and display the calculated centroid of the figure. The measuring device comprises a main body having a roller capable of running in a single direction and a measuring lever pivotably supported by the main body so as to be rotatable right or left from the running direction of main body on a horizontal plane. In the measuring device, the main body has a display and a calculating circuit. The tracer of the measuring lever traces the contour of the figure to measure the coordinates, areas, and lengths of the figure. At the same time, the calculating circuit computes the geometrical moments of areas successively and sums the geometrical moments of the areas. Finally, the calculating circuit divides the summed geometrical moment of the areas by an area of the figure to obtain the coordinates of the centroid of the figure.

Abstract: A targeting system which receives input data from two sources having different video characteristics. The apparatus includes a multiple-pass centroid eroder, which operates at a multiple of the video input rate. The apparatus identifies "hot spots," i.e., potential targets, and uses the identified potential targets from the two video sources to correlate targeting of submodules producing the two video signals.

Abstract: A motion analysis system utilizing an image processing technique includes a color marker attached to an object which is to be analyzed. Each color marker has a high directivity reflective member on a surface thereof, and the reflection member reflects light having a particular wavelength. A light source generates light for irradiation onto the color marker. A color TV camera aligned in roughly the same direction as that of the light source receives reflected light from the color marker and outputs a color image. A color extraction unit connected to the TV camera, extracts only a particular color from the color image produced by the TV camera. An area calculation unit connected to the color extraction unit, calculates the area of the particular color extracted by the color extraction unit.

Abstract: A one or two player virtual reality game efficiently detects and tracks a distinctively colored glove. According to the preferred basketball embodiment, a single player equipped with the distinctively colored glove is matched up against a virtual opponent. The object of the game is for the real player to put a virtual basketball into a virtual basketball hoop before his/her virtual opponent steals the ball. Initially, the background site is scanned, and then the operator with the glove is scanned. A table of colors is then established which are unique only to the glove. A player is then scanned against the background to identify which color glove will have the least conflict with colors worn by the player. During play, the player is scanned at 30 frames a second and the information is stored in a frame buffer. A prediction is made of the location of the glove in subsequent frames based upon its previously known location and velocity.

Abstract: A method and device for use in detecting moving targets in a scene represented by a sequence of mutually registered images R ( . . . , R.sub.n-1, R.sub.n, R.sub.n+1, . . . ) over time wherein differential images D ( . . . , D.sub.n-1, D.sub.n, D.sub.n+1, . . . ) are derived from succeeding images R, and spatially low pass filtered to obtain a sequence of filtered differential images F ( . . . , F.sub.n-1, F.sub.n, D.sub.n+1, . . . ). For each pixel of a plurality of pixels (i,j) and for each of a plurality of directions (d.sub.x,d.sub.y), a sum of intensities is formed from a number of successive frames of the sequence of filtered differential images F.sub.n-m (m=0,1 . . . M) at pixel locations (i+md.sub.x,j+md.sub.y) located, as the image is further back in time, along the direction progressively outwards from the pixel.

Abstract: A moving object tracking method for determining whether a moving object has come to a standstill and whether the moving object is moving irregularly without defining a background image. Images of a predetermined area are input sequentially. As the images are input, an object moving within the predetermined area is extracted based on the differences between the images. A representative point of each such moving object is calculated for each new image, and the expected range of movement of the representative point is determined. The moving object is found to be moving regularly when the representative point remains within the expected range.