Data recognition Library

Key Requirements 1. Get text from an image (Given a region). High accuracy 98% or better, dirty or complex backgrounds 2. Patron comparison (Detect and calculate similarity) ** Compatible with Windows and Linux Additional requirements (Negotiable) -. Method to get differential image. -. Method to compare two images (different sizes) -. Method that load the result file and return all the cuts (Regions) -. Pre-processing: Correct possible rotation. -. Wrapper for .Net Expected Flow 1. Load the configuration data from JSON or XML Object (File or memory). (Have to be an application parameter) 2. Load the inspected image 3. For each Text Region 3.1 Clean the image region 3.2 Get the text inside de region 3.3 The result will be stored in the result document. 4. For each Image Region 4.1 without sample imagen 4.1.1 Store the new coordinates and size 4.2 with sample imagen 4.2.1 Compare the region and the image 4.2.2 Store the new coordinates and size 4.2.3 Store the similirity 5. Result Object in JSON or XML format (Memory or File) ** If the inspected image has a different size, all the coordinates will Configuration Content 1. Expected image size and resolution (Sample size) 2. List of text regions A text region is defined by an unique name or identifier and the coordinates where is expected the text (X,Y). ** possible additional configuration for the image processing or the ocr process, for example, the font to process (Proposed by the developer) 3. List of image regions A image region is defined by an unique name or identifier, the coordinates where is expected the image (X,Y). and optional image (Base 64) for comparison purpose. Required OCR Fonts: - Arial - OCR A Deliverable: - Compiled DLL - Compiled SO (Optional) - Use example (Command Line or Simple UI) - Install and Use instructions Related technologies: - C++ (Dynamic Library) - OpenCV - Tesseract - QT (Optional)