This paper provides a survey of methods and strategies in character segmentation, focusing on the techniques used in optical character recognition (OCR). Character segmentation is a critical step in OCR, as it involves decomposing an image of a sequence of characters into subimages of individual symbols. The paper discusses three main strategies for character segmentation: the classical approach, recognition-based segmentation, and holistic methods. The classical approach involves dissection, which is the decomposition of the image into classifiable units. Recognition-based segmentation uses classification to select from possible segmentation possibilities, while holistic methods recognize entire character strings as units, avoiding the need for segmentation into characters. The paper also discusses the challenges of character segmentation, including the interdependence of segmentation decisions with local and global decisions. It highlights the limitations of the classical approach and the need for more sophisticated methods that integrate segmentation and classification. The paper reviews various techniques for segmentation, including dissection, projection analysis, connected component processing, and contextual postprocessing. It also discusses the use of Hidden Markov Models (HMMs) and other statistical methods in character segmentation. The paper concludes that while there are many approaches to character segmentation, the most effective methods often combine different strategies to achieve accurate and reliable results.This paper provides a survey of methods and strategies in character segmentation, focusing on the techniques used in optical character recognition (OCR). Character segmentation is a critical step in OCR, as it involves decomposing an image of a sequence of characters into subimages of individual symbols. The paper discusses three main strategies for character segmentation: the classical approach, recognition-based segmentation, and holistic methods. The classical approach involves dissection, which is the decomposition of the image into classifiable units. Recognition-based segmentation uses classification to select from possible segmentation possibilities, while holistic methods recognize entire character strings as units, avoiding the need for segmentation into characters. The paper also discusses the challenges of character segmentation, including the interdependence of segmentation decisions with local and global decisions. It highlights the limitations of the classical approach and the need for more sophisticated methods that integrate segmentation and classification. The paper reviews various techniques for segmentation, including dissection, projection analysis, connected component processing, and contextual postprocessing. It also discusses the use of Hidden Markov Models (HMMs) and other statistical methods in character segmentation. The paper concludes that while there are many approaches to character segmentation, the most effective methods often combine different strategies to achieve accurate and reliable results.