CHARACTERIZATION OF VIBRATION-INDUCED IMAGE DEFECTS IN INPUT SCANNERS Robert P. Loce George Wolberg Xerox Corporation Department of Computer Science Corporate Research and Technology City College of New York / CUNY Webster, NY 14580 New York, NY 10031 loce.wbst147@xerox.com wolberg@cs-mail.engr.ccny.cuny.edu ABSTRACT Typical office scanners employ a moving linear-sensor array or moving optics. the velocity of the components is gen- erally not constant in time. It may be modulated directly (at one or more frequencies) by dynamic errors of gears, timing bets, and motors, and indirectly by structural vibrations induced by gears, fans, etc. Nonuniform velocity is known to cause undesirable brightness fluctuation and warping in the sampled image. The present paper characterizes the image defects induced by nonuniform velocity. A companion paper utilizes the degradation information to develop an algorithm to restore the degraded image. Keywords: image defects, digital documents, scanned documents, vibrations, motion quality 1. INTRODUCTION In a document scanner, a linear sensor array (LSA) passes over a document, ideally, at uniform velocity. The output of a single photosensor in the LSA is proportional to the integrated irradiance received by the sensor as it scans through an exposure region. When vibratory motion of the sensor array is considered, the actual size of the exposure region is time- varying. This introduces artifacts such as brightness fluctuation and geometric warping. In this paper, we characterize these image defects. The effect of motion on the resolution of analog imaging systems has been studied extensively. The problem typically addressed concerns characterization of the image blur caused by motion during the integration time of a photosensitive imag- ing medium. In this typical case the blur is usually constant throught the image and may be caused by linear or vibratory motion. The blur is constant because the devices under study acquire the full spatial extent of the image at one time, as in a conventional photographic camera. Characterization of this spatial uniform blur leads to a deconvolution-type image restora- tion algorithm. The present paper differs from that body of work in that we are concerned with images that are digitized in a scan line fashion where each scan line may be acquired with a different arbitrary velocity of the light sensing array. For more information on the uniform blur case see [1, 2, 3]. Several authors have considered the effects of nonuniform photoreceptor motion and other sources of noise in slit scanning photocopiers and digital printers. Loce and Lama [4] characterized the periodic exposure level error caused by vibration in a slit scanning photocopier. Bestenreiner, Geis, Helmberger, and Stadler [5], Takiguchi, Miyagi, Okamura, Ishoshi, and Shibata [6], and Haas [7] examined the effects of periodic scan line position errors when printing periodic binary patterns (e.g., halftones). Schubert [8] and Firth, Kessler, Muka, Noar and Owens [9] analyzed banding in continuous tone prints due to periodic errors. Burns, Rabbani and Ray [10], and Melnychuck and Shaw [11] concentrated on the effect of random errors in the continuous tone and binary cases, respectively. Bloomberg and Engeldrum [12] analyzed the color error on a print that is caused by random pixel placement errors. Loce and Lama [13] employed exposure and xerographic models to examine vibration induced halftone banding in image bar printers. Loce, Lama and Maltz [14] examined the effects of raster position error in a xerographic laser printer and focused on the parametric dependences of halftone banding in the resulting prints. This paper describes the basic operation of a moving photosensor and the problem of vibrations in input scanners. An analysis and examples of image defects for several special cases is given. A companion paper [16] utilizes the degradation information to develop an algorithm that restores the degraded image.