OCR Integration
Latest stable Based on Image release 0.3.0
bluetape4k-images-ocr adds Tesseract text extraction to the core ImmutableImage model. It is a separate module because OCR brings native packages, language data, latency, and operational failure modes that normal image processing does not need.
API model
Section titled “API model”OcrOptions selects language and engine behavior. OcrResult carries recognized text and result data. TesseractOcrEngine implements the engine, and ImmutableImage.extractText is the convenient entry point for an existing image value.
Configure language and data path explicitly in controlled deployments. Normalize or resize input only when it improves the document class being processed; aggressive filters can remove characters as easily as noise.
Service design
Section titled “Service design”Bound upload bytes, decoded dimensions, OCR concurrency, and request time. OCR is a good candidate for a worker queue when documents are large or latency is unpredictable. Apply the same data-protection and retention policy to original media and extracted text; recognized text can be more searchable than the image.
Distinguish configuration failure, engine failure, empty recognized text, and request validation. Do not retry malformed or unsupported input indefinitely.
Learn with runnable paths
Section titled “Learn with runnable paths”Complete OCR setup first. Then run the Ktor OCR or Spring Boot OCR workshop and read its tests with the application code.