Currently researchers are attracted to the area of Optical Characterrnrecognition primarily due to challenging nature of the research and secondlyrndue to the industrial importance that it provides in the area of Readingrnmachine for the Blind, postal Address interpretation, Bank Curtsey amountrnprocessing, hand filled form processing, and the like.rnResearch III the area of Amharic OCR systems is ongoing Since 1997.rnAttempts were made III adopting algorithm to Amharic language,rnincorporating preprocessing techniques to the adopted algorithm, and inrngeneralizing the system so as it recognizes Type written characters a::; well asrnhand written charactersrnSufficient amount of work is done in the areas of preprocessing such asrnsegmentation and Noise Removal. However, the consideration given to thernsimplification of the feature extraction and the efforts made to alleviate thernproblems of high dimensional input still requires the contribution of manyrnadditional researches in order to come up with a system that the society canrnuse to solve real world problems.rnTo this end, Line fitting is used to Amharic Optical character recognition byrnapplying simple geometric calculations to determine features which couldrnrepresent and describe the character as uniquely and precisely as possible.rnThe image of a segmented character which is normalized into 32x32 pixels isrndivided into 16 smaller squares of 8x8 pixels. Then the least square techniquernwas applied to fit a linear model to the distribution of foreground pixels andrnthree features were extracted from each smaller squarernFinally, a feed forward Neural Network trained using a back propagationrnalgorithm is used on handwriting of three individuals using a cross validationrntechnique as well as a separate test set and results are depicted on tables andrnconfusion matricesrnRelevant Conclusions were drawn and some valid recommendations werernforwarded to indicate future direction of further works on the area.rncombining the shape of the letters so as to form written words) [Plamondonrnand Srihari, 2000].rnHandwriting, since it entails an individualistic skill and contains artificialrngraphical marks on the surface, is still a challenge in pattern recognition. Thernsuccess of handwritten optical character recognition system is attributed tornthe availability of machine learning techniques [Lecun et.al 1998]. However,rnthe availability of machine learning techniques alone is not able to solve thernproblems of offline OCR systems. To this end, some of the problems remainrnrather far away from being solved successfully.rnSince 1951, a time remarked by the invention of GISMO - a robot readerrnwriter, many OCR systems were developed due to the advantages that theyrnprovide in overcoming the problem of repetitive and labor intensive tasksrn[Srihari & Lam, 1996]. At present hundreds of OCR systems arerncommercially available, and they are less expensive, faster , and more reliablerndue to less expensive electronic components, and extensive researches in thernarea [Yare gal, 2002].rnTechnically, Handwriting Recognition Systems compnse procedures likernScanning documents, Binarization, segmentation, feature extraction,rnrecognition, and/or possible post processing [Million, 2000; De Lesa, 2001].rnAs Dereje mentioned in 1999, the OCR systems are highly influenced byrnfactors like mode of writing, condition of the input, quality of the paper, andrnthe presence of extraneous marks. In order to increase the performance ofrnOCR systems, various preprocessing tasks like noise removal, skew detectionrnand correction, and slant correction were applied to printed and type writtenrnscripts. Effort was also made in using structural features partly to increasernthe versatility of OCR systems [Yare gal, 2002) .rnIn addition to the problems of machine printed and type written scripts,rnhandwriting recognition has additional inconveniences introduced becausernof the great inconsistency of writing styles, and handwriting instruments.