Showing papers on "Bessel filter published in 1975"

Video traveling matte system

Abstract: A video traveling matte system is disclosed in which the video signal from a foreground television camera is examined for the presence of a particular color by an arrangement employing multipliers to multiply the differences between the red and blue components of the video signal and a luminance signal derived from the video signal by the signals from a sine/cosine potentiometer. The combined signal from the outputs of the multipliers is fed to a first comparator for comparison with a reference value, and the resulting pulses are filtered by a 4-pole active Bessel filter and compared with a reference value in a second comparator to increase the width of the pulses by a selected amount. The pulses of increased width prevent the formation of a matte outline between foreground and background images as the true and complementary values thereof are applied to alternately gate the video signals from the foreground television camera and a background television camera to a common output. The effects of shadows in the foreground image are maintained despite substitution of the background image by utilizing the combined signal from the multipliers to modulate pulses from the second comparator prior to use of such pulses to gate the video signal from the background television camera to the common output. The effects of shadows are further enhanced by a logarithmic amplifier which introduces a predetermined logarithmic transfer characteristic into the modulation signal to compensate for the black stretch circuits used in most television cameras.

A modification of the Bessel filter

Abstract: The continued fraction used to obtain the transfer function of the Bessel filter is modified by adding a parameter a > -1 to the odd integers in the expansion. The resulting transfer function is realizable and has a time delay at \omega = 0 which is independent of the order of the filter, and which may be varied from 0 to \infty by changing a . A further modification yields transfer functions from continued fraction expansions involving the even integers and the positive integers.

Remark on “Algorithm 236: Bessel Functions of the First Kind [S17]”

Abstract: The procedures in Algorithm 236 were coded in PL / I and run on the IBM 370/165. The following error was discovered for a = O, nmax large, and x small, e.g. nmax = 50 and x = 0.5. In the last if statement in three of the procedures, Japlusn, Iaplusn, and Complex Japlusn, division by zero took place. Not all compilers and computers would pose problems for the above values of the parameters; whether or not they do depends on the permissible magnitude of the flouting-point numbers for the compiler and computer used. For the IBM 370/165 the smallest positive floating-point number which the computer can hold is approximately 5.40 X 10 -79 (see [10, p. 163]). The following corrections should be made in the procedure Japlusn. The last if statement should be replaced by