This document presents a new algorithm called Skip Trie Matching (STM) for correcting errors from optical character recognition (OCR) outputs on smartphones in real time. STM uses a trie data structure containing a dictionary of strings to correct errors by skipping misrecognized characters as it searches the trie. It is compared to existing spell checking algorithms like n-gram matching and Levenshtein edit distance. The document provides context by describing the prior steps in a mobile vision system for extracting nutrition information from product packages, including locating nutrition labels and segmenting text. It then details the STM algorithm and analyzes its performance compared to other methods.