Bankcheck Recognition using Cross Validation Between Legal and Courtesy Amounts

Abstract

A bankcheck reading system using cross validation of both the legal and the courtesy amounts is presented in this paper. Some of the challenges posed by the task are (i) segmentation of the legal amount into words, (ii) location of boundaries between dollars and cents amounts, and (iii) high accuracy in terms of recognition performance. Word segmentation in the legal amount is a serious issue because of the nature of the data and patrons' writing habits which tend to clump words together. We have developed a word segmentation algorithm based on the character segmentation results to address this issue. The list of possible amounts generated by the word segmentation hypotheses is used as lexicon for the courtesy amount recognition. The order of magnitude of the amount is estimated during legal amount recognition. We treat the courtesy amount as a numeral string and apply the same word recognition scheme as used for the legal amount. Our approach to check recognition differs from traditional methods in two significant aspects: First, our emphasis on both the legal and the courtesy amounts is balanced. We use an accurate word recognizer which performs equally well on alpha words and digit strings. Second, our combination strategy is serial rather than the commonly used parallel method. Experimental results show that 43.8% of check images are correctly read with an error rate of 0%.

The role of holistic paradigms in handwritten word recognition

“…Several examples of holistic recognizers that work with dynamic lexicons have been developed for use in the postal handwritten address interpretation context and described in the literature [11], [22], [23], [34]. Most systems used in the check recognition application [28] work with static lexicons.…”

Automatic recognition of handwritten medical forms for search engines

“…Although handwriting recognition and lexicon pruning/reduction [33] have been researched substantially over the years, many challenges still persist in the offline domain. Word recognition applications range from automated check recognition [26], postal recognition [14], historical documents recognition [15] [18], and now emergency medical documents [35] Lexicon reduction has been shown to be critical to improvement of performance primarily because of the minimization of possible choices [19]. Even the systems dealing with a large vocabulary corpus have been successful [28] [29].…”

Segmentation and recognition of Chinese bank check amounts

“…Westall, H. Su [Westall 1997, Su 1996]. Cross validation of courtesy and legal amounts is also a hot topic [Dzuba 1997, Kim 1997]. Some complete systems have…”