@techreport{LAMP_030,
    author = { {T}. {K}anungo and {G}. {M}arton and {O}. {B}ulbul },
    abstract = { {C}haracterizing the performance of {O}ptical {C}haracter {R}ecognition ({OCR}) systems is crucial for monitoring technical progress, predicting {OCR} performance, providing scientific explanations for system behavior and identifying open problems. {W}hile research has been done in the past to compare the performances of {OCR} systems, all methods assume that the accuracies achieved on individual documents in a dataset are independent. {I}n this paper we argue that accuracies reported on any dataset are not independent and invoke the appropriate statistical technique -- the paired model -- to compare the accuracies of two recognition systems. {W}e show theoretically that this method provides tighter confidence intervals than the methods used in the {OCR} and computer vision literature. {W}e also propose a new visualization method, which we call the accuracy scatter plot, for providing a visual summary of performance results. {T}his method summarizes the accuracy comparisons on the entire corpus while simultaneously allowing the researcher to visually compare the performances on individual document images. {F}inally, we report on the accuracy and speed performances as functions of image resolution. {C}ontrary to what one might expect, the performance of one of the systems degrades when the image resolution is increased beyond 300 dpi. {F}urthermore, the average time taken to {OCR} a document image, after increasing almost linearly as a function of resolution, suddenly becomes a constant beyond 400 dpi. {T}his behavior is most likely because the {S}akhr {OCR} algorithm resamples the high-resolution images to a standard resolution. {T}he two products that we compare are the {A}rabic {O}mni{P}age 2.0 and the {A}utomatic {P}age {R}eader 3.01 from {S}akhr. {T}he {SAIC} {A}rabic dataset was used for the evaluations. {T}he statistical and visualization methods presented in this paper are very general and can be used for comparing the accuracies of any two recognition systems, not just {OCR} systems. },
    institution = { {U}niversity of {M}aryland, {C}ollege {P}ark },
    month = { {D}ecember },
    number = { {LAMP}-{TR}-030,{CFAR}-{TR}-903,{CS}-{TR}-3972 },
    pdffile = { http://lampsrv02.umiacs.umd.edu/pubs/TechReports/LAMP_030/LAMP_030.pdf },
    psfile = { http://lampsrv02.umiacs.umd.edu/pubs/TechReports/LAMP_030/LAMP_030.pdf },
    title = { {P}aired {M}odel {E}valuation of {OCR} {A}lgorithms },
    year = { 1998 }
}