# ub-mannheim/tesseract architecture

> Architecture diagram of the GitHub repository ub-mannheim/tesseract, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-01-29). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/ub-mannheim/tesseract
- Repository: https://github.com/ub-mannheim/tesseract

## Overview

##### Main Components

a. Core Engine Components:
- API Layer (src/api/)
- LSTM Neural Network Engine (src/lstm/)
- Legacy OCR Engine
- Text Ordering System (src/textord/)
- Word Recognition System (src/wordrec/)
- Classification System (src/classify/)
- Dictionary System (src/dict/)

b. Supporting Components:
- Training System (src/training/)
- Image Processing Pipeline
- Data Management System (tessdata/)
- Viewer/Debug Interface (src/viewer/)

##### Component Relationships and Data Flow

**Input Image → Image Processing → Page Layout Analysis → Line Detection → Word Detection → Character Recognition → Post-processing → Output**

##### Detailed Component Breakdown

a. Input/Output Layer:
- Input formats (PNG, JPEG, TIFF)
- Output formats (Plain text, hOCR, PDF, TSV, ALTO)
- API interfaces (C/C++)

b. Recognition Pipeline:
- Image preprocessing
- Layout analysis
- Line finding
- Word segmentation
- Character recognition
- Language modeling

##### Architecture Patterns

Following architectural elements:
- Modular design with clear component separation
- Pipeline processing structure
- Plugin architecture for language data
- Multi-engine support (LSTM and Legacy)

##### Key Technical Elements

Specialized components:
- LSTM neural network architecture
- Language model integration
- Unicode support system
- Training infrastructure

- Applications  |

- API Layer    |

- Core Engines   |

- Support Systems|

##### Integration Points

Connections with:
- Leptonica library integration
- External application interfaces
- Training tools and data
- Language data files

## Diagram (Mermaid)

```mermaid
graph TB
    %% Style definitions
    classDef core fill:#2374f7,color:white
    classDef io fill:#4CAF50,color:white
    classDef support fill:#FFC107,color:black
    classDef critical fill:#f44336,color:white
    classDef api fill:#9C27B0,color:white

    subgraph Input
        InputImage["Input Image\n(PNG, JPEG, TIFF)"]:::io
    end

    subgraph "API and Control Layer"
        API["API Layer"]:::api
        MainControl["Main Control"]:::core
    end

    subgraph "Core Processing"
        ImageProc["Image Processing"]:::core
        CoreUtils["Core Utilities"]:::core
        CoreStruct["Core Structure"]:::core
        ArchOpt["Architecture Optimizations"]:::core
    end

    subgraph "Recognition Engines"
        LSTM["LSTM Neural Network"]:::core
        Legacy["Legacy Engine"]:::core
    end

    subgraph "Text Analysis"
        TextOrd["Text Ordering"]:::core
        WordRec["Word Recognition"]:::core
        Classify["Classification"]:::core
        Dict["Dictionary"]:::core
    end

    subgraph "Support Systems"
        Training["Training System"]:::support
        Viewer["Viewer/Debug Interface"]:::support
        DataMgmt["Data Management"]:::support
        Testing["Unit Testing"]:::support
    end

    subgraph Output
        OutputText["Output\n(Text, hOCR, PDF, TSV)"]:::io
    end

    %% Connections
    InputImage --> API
    API --> MainControl
    MainControl --> ImageProc
    ImageProc --> CoreUtils
    CoreUtils --> CoreStruct
    CoreStruct --> |Layout Analysis| TextOrd
    TextOrd --> |Word Detection| WordRec
    WordRec --> |Character Recognition| Classify
    Classify --> Dict
    Dict --> OutputText

    MainControl --> LSTM
    MainControl --> Legacy
    LSTM --> TextOrd
    Legacy --> TextOrd

    DataMgmt --> Dict
    DataMgmt --> Training
    Training --> LSTM
    Viewer --> MainControl
    ArchOpt --> MainControl
    Testing --> MainControl

    %% Click events
    click API "https://github.com/UB-Mannheim/tesseract/tree/main/src/api/"
    click LSTM "https://github.com/UB-Mannheim/tesseract/tree/main/src/lstm/"
    click TextOrd "https://github.com/UB-Mannheim/tesseract/tree/main/src/textord/"
    click WordRec "https://github.com/UB-Mannheim/tesseract/tree/main/src/wordrec/"
    click Classify "https://github.com/UB-Mannheim/tesseract/tree/main/src/classify/"
    click Dict "https://github.com/UB-Mannheim/tesseract/tree/main/src/dict/"
    click Training "https://github.com/UB-Mannheim/tesseract/tree/main/src/training/"
    click Viewer "https://github.com/UB-Mannheim/tesseract/tree/main/src/viewer/"
    click DataMgmt "https://github.com/UB-Mannheim/tesseract/tree/main/tessdata/"
    click CoreUtils "https://github.com/UB-Mannheim/tesseract/tree/main/src/ccutil/"
    click CoreStruct "https://github.com/UB-Mannheim/tesseract/tree/main/src/ccstruct/"
    click MainControl "https://github.com/UB-Mannheim/tesseract/tree/main/src/ccmain/"
    click ArchOpt "https://github.com/UB-Mannheim/tesseract/tree/main/src/arch/"
    click Testing "https://github.com/UB-Mannheim/tesseract/tree/main/unittest/"

    %% Legend
    subgraph Legend
        L1["Core Component"]:::core
        L2["I/O Component"]:::io
        L3["Support System"]:::support
        L4["Critical Path"]:::critical
        L5["API Component"]:::api
    end
```
