Evaluation of Character Recognition Algorithm Based on Feature Extraction Bishakha Sharma Research scholar, Deptt. Of Computer Science & Engg. ITM Group Of Institution,Gwalior (M.P) bishakhasharmajmp@gmail.com Arun Agarwal Assistant Professor, Deptt. Of Computer Science & Engg ITM Group Of Institution, Gwalior (M.P.) arun.agarwal@itmgoi.in Abstract-At display circumstance, there is creating enthusiasm for the item system to see characters in a PC structure when information is investigated paper records. This paper presents point by point review in the field of Optical Character Recognition. Diverse methods are settled that have been proposed to comprehend the point of convergence of character affirmation in an optical character affirmation structure. Decision and feature extraction in light of Optical Character Recognition (OCR). By using the OCR, we can change the information of picture into the information of substance which is definitely not hard to control. In our proposed method, Select the any particular number and crop the selected image and then extract the feature. The text from the OCR process will be compared with the selected number from the loaded image. The overall accuracy of the proposed method is 92%. . Keywords—image processing; Optical Character Recognition; Feature Extraction; I. INTRODUCTION A piece of software through which printed text and images can be converted into digitized form such that it can be manipulated by machine is known as character recognition system.The human brain which has the capability to very easily recognize the text/ characters from an image, but machines have not enough capability to perceive image information. Therefore, a large number of research efforts have been put forward that attempts to transform a document image to format understandable for machine. OCR is a mind boggling issue in light of the assortment of dialects, scholarly styles and styles in which substance can be made, and the versatile tenets of tongues and so on. Thusly, frameworks from various solicitations of programming outlining (i.e. picture dealing with, design depiction and trademark vernacular arranging and so forth are utilized to address grouped difficulties. This paper acclimates the peruser with the issue.It enlightens the reader with the historical perspectives, applications, challenges and techniques of OCR. [1] Optical Character acknowledgment has been a subject of research. Example acknowledgment has three fundamental advances: perception, design division, and example arrangement. Optical Character Recognition (OCR) frameworks is changing substantial measure of records, either printed letters in order or transcribed into machine encoded content with no change, tumult, affirmation blends and different segments. At the point when all is said in done, handwriting affirmation is portrayed into two sorts as offline and On-line character recognition. Offline recognition includes programmed change of content into a picture into letter codes which are usable inside PC and content preparing applications. Offline recognition is more troublesome, as different people have assorted handwritten styles. Be that as it may, in the on- line framework, On-line character acknowledgment manages an information stream which originates from a transducer while the client is composing. The normal equipment to gather information is a digitizing tablet which is electromagnetic or weight delicate. At the point when the client composes on the tablet, the progressive developments of the pen are changed to a progression of electronic flag which is remembered and investigated by the PC. Optical Character Recognition (OCR) is a field of research in design acknowledgment, manmade brainpower and machine vision,signal processing. It is additionally aforementioned that Optical character recognition(OCR) is reffered to as associate Off-line character recognition system during which system scans and static image of the characters ought to be recognized.It alludes to the mechanical or electronic interpretation of pictures of manually written character or printed content into machine code with no