Final Call for Papers for a Special Issue of 
             Computer Vision and Image Understanding 
                            on
       Empirical Evaluation of Computer Vision Algorithms 
       http://www.itl.nist.gov/div894/vip/cviu/cviu_cfp.html
 
 Computer vision has advanced to the point where a stronger experimental
 tradition is needed.  As the importance of this need becomes better
 recognized, more researchers are focusing effort on sound experimental
 work.  To promote the development of a stronger experimental tradition
 in computer vision and to inform the computer vision community of
 progress in this area, we are collecting papers on this subject for a
 special issue. 
 
 We are soliciting papers on ``Empirical Evaluation of Computer Vision
 Algorithms'' to appear in a special issue of Computer Vision and Image
 Understanding (CVIU).  Guest editors for the special issue are Patrick
 Flynn of Ohio State University, Adam Hoover of University of California,
 San Diego, and Jonathon Phillips of the National Institute of Standards
 and Technology. 
 
 Papers appropriate to this issue will have empirical evaluation of
 vision algorithms as their central focus, with the evaluation cutting
 across all algorithmic areas of computer vision.  This includes optic
 flow, edge detection, range image segmentation, face and gesture
 recognition, document image analysis, image database retrieval, 3-D
 model construction, motion tracking, document processing, or any other
 topic area that is appropriate to CVIU. 
 
 Ideally, papers will define and demonstrate empirical evaluation
 techniques that set standards for various sub-areas in computer vision.
 These standards should serve as reference points for future work and
 publications.  Types of papers that are of interest would define a
 convincing performance evaluation methodology for a class of computer
 vision algorithms, evaluate a set of well-known algorithms using this
 methodology, assess the state-of-the-art in for this class of algorithms
 provide the strengths and weakness of each algorithm, and provide a set
 of software tools, images, and associated resources for future
 evaluations. 
 
 Also of interest are papers that provide a sound statistical methodology
 for analyzing performance results.  Papers that evaluate a single
 algorithm for a given problem may be of interest, if the tools and
 methods used are readily adapted for use by other researchers.  Papers
 that are primarily conceptual or theoretical are less likely to be
 deemed appropriate. 
 
 All submitted papers will be reviewed according to the guidelines and
 standards of CVIU. Four copies of submitted papers should be sent to:
   Dr. Jonathon Phillips, NIST, Bldg 225 / Rm A216, 
   100 Bureau Dr Stop 8940, Gaithersburg, MD 20899-8940 
 
 Manuscripts are to be submitted by 15 September 1999.  First reviews
 should be returned to authors December 1999.  Revised manuscripts will
 be due by the end of 15 February 2000.  The special issue should appear
 in July or August 2000.