Integrating Tess4j with SpringBoot: Low‑Cost OCR Image Recognition

Overview

SpringBoot can expose OCR functionality through a lightweight REST service by using tess4j, a Java wrapper for the Tesseract engine. The service processes image files and returns the recognized text.

Demo

The expected OCR result is shown in the following image.

Implementation

1. Maven Dependency

<dependency>
  <groupId>net.sourceforge.tess4j</groupId>
  <artifactId>tess4j</artifactId>
</dependency>

2. Tesseract Engine Initialization

When the application is packaged as a JAR,

new ClassPathResource("tess_data").getFile().getAbsolutePath()

may fail. The article recommends copying the resource to a temporary location (e.g., via a TensorflowUtil helper) before passing the path to Tesseract.

On Linux, the native library net.sourceforge.tess4j.TessAPI must be loadable; all required system dependencies need to be installed.

3. Service Class

public class TesseractOcrModelService {
    private final Tesseract tesseract = new Tesseract();
    public TesseractOcrModelService() {
        try {
            String folderPath = new ClassPathResource("tess_data").getFile().getAbsolutePath();
            // OEM_TESSERACT_LSTM_COMBINED = 2 – run both Tesseract and LSTM, fallback to Tesseract when needed
            tesseract.setPageSegMode(OEM_TESSERACT_LSTM_COMBINED);
            tesseract.setDatapath(folderPath);
            // Page segmentation mode 6 – assume a single uniform block of text
            tesseract.setPageSegMode(6);
            tesseract.setLanguage("chi_sim"); // Simplified Chinese
        } catch (Exception e) {
            throw new RuntimeException(e);
        }
    }
    public Tesseract getTesseract() { return tesseract; }
}

4. RESTful Controller

@RestController
@RequestMapping("ocr")
@RequiredArgsConstructor
public class OcrController {
    private final TesseractOcrModelService tesseractOcrModelService;

    @PostMapping("/detection")
    public Result<String> ocrDetection(MultipartFile file) {
        try {
            Tesseract tesseract = tesseractOcrModelService.getTesseract();
            // Recommended preprocessing: binarization, denoising, rotation correction (not implemented here)
            return Result.success(tesseract.doOCR(ImageIO.read(file.getInputStream())));
        } catch (Exception e) {
            throw new RuntimeException("ImageIO.read(file.getInputStream()) parsing error");
        }
    }
}

5. Training Data Options

tessdata_best

– highest accuracy, slower recognition. tessdata – balanced speed and accuracy, suitable for general OCR. tessdata_fast – optimized for speed, lower accuracy.

Source Code

https://gitee.com/fateyifei/yf

Conclusion

Tess4j reliably extracts ID numbers, phone numbers, and English words. Using the free Chinese models ( chi_sim) yields lower accuracy for Chinese text. Higher quality can be achieved by training custom data sets or by invoking third‑party OCR APIs such as Google Cloud Vision, Microsoft Azure OCR, or Amazon Textract.

Potential application scenarios include document digitization, automatic data entry from scanned tables, license‑plate recognition, and handwritten text conversion.