Rule-Based Systems

The earliest OCSR systems converted raster images into vector primitives (lines, arcs, characters) and applied hand-coded chemical rules to assemble those primitives into molecular graphs. Pioneering tools like Kekulé (1992), CLiDE (1993), and the Contreras system (1990) established the core pipeline: binarize, thin, vectorize, classify atoms and bonds, then compile a connection table. Later systems such as OSRA, ChemReader, and CLiDE Pro refined each stage with better segmentation, chemical spell-checking, and support for superatom labels. The approach dominated the field for over two decades, but brittleness in the face of diverse drawing styles, noisy scans, and edge-case notation ultimately motivated the shift to learned representations.

Year	Paper	Key Idea
1990	Graph Perception for Chemical Structure OCR	Early OCR system for digitizing chemical structures using C and Prolog
1992	Kekulé: OCR-Optical Chemical Recognition	Seminal OCSR system using neural networks and heuristic graph compilation
1993	Chemical Literature Data Extraction: The CLiDE Project	Seminal system converting scanned diagrams into connection tables
1993	Optical Recognition of Chemical Graphics	Prototype using vectorization and heuristic-based structure recognition
1996	Kekulé-1 System for Chemical Structure Recognition	Neural OCR with chemical rule-based post-processing for OCSR
2003	Chemical Machine Vision	Gabor wavelets and Kohonen networks for chemical image classification
2007	Automatic Recognition of Chemical Images	Rule-based extraction validated against commercial baselines
2007	Reconstruction of Chemical Molecules from Images	5-module system converting raster images to SDF with custom vectorization
2009	ChemReader: Automated Structure Extraction	Modified Hough transform and chemical spell checking for OCSR
2009	CLiDE Pro: Optical Chemical Structure Recognition Tool	OCSR system reconstructing chemical graphs with ~90% accuracy
2009	OSRA: Open Source Optical Structure Recognition	First open-source utility for converting chemical images to SMILES/SD
2011	ChemInfty: Chemical Structure Recognition in Patent Images	Segment-based approach for challenging Japanese patent images
2012	Chemical Structure Recognition (Rule-Based)	Rule-based expert system (MolRec) using 18 geometric rewrite rules
2014	Probabilistic OCSR with Markov Logic Networks	Markov Logic Networks replacing brittle heuristics for robustness
2015	Research on Chemical Expression Images Recognition	Improved handling of adhesive symbols and wedge bonds
2026	GraphReco: Probabilistic Structure Recognition (2026)	Markov networks for probabilistic atom/bond ambiguity resolution

Computational Chemistry

Early optical recognition system converts scanned chemical diagrams to connection tables

Optical Recognition of Chemical Graphics

This paper describes an early prototype system that digitizes chemical structure diagrams from scanned documents. It employs a multi-stage pipeline involving convex bounding polygon extraction, vectorization, and rule-based heuristics to generate MDL Molfiles.

Computational Chemistry

Chemical structure diagram for optical recognition

OSRA: Open Source Optical Structure Recognition

This paper presents OSRA, the first open-source utility for converting graphical chemical structures from documents into machine-readable formats (SMILES/SD). It outlines a pipeline combining existing image processing tools with custom heuristics for bond and atom detection, establishing a foundation for accessible chemical information extraction.

Computational Chemistry

Five-stage pipeline for reconstructing chemical molecules from raster images

Reconstruction of Chemical Molecules from Images

This methodological paper proposes a comprehensive pipeline to digitize chemical structure images. It achieves 97% reconstruction accuracy on benchmarks by combining a topology-preserving vectorizer with a chemical knowledge validation module.

Computational Chemistry

Segment-based chemical structure recognition pipeline for low-quality patent images with touching characters and broken lines

ChemInfty: Chemical Structure Recognition in Patent Images

A 2011 rule-based OCSR system designed specifically for the challenging low-quality images in Japanese patent applications, using segment-based methods to handle pervasive problems like touching characters, merged atom labels with bonds, and broken lines.