OpenCV
3.2.0
Open Source Computer Vision
|
Classes | |
class | cv::text::BaseOCR |
class | cv::text::OCRHMMDecoder |
OCRHMMDecoder class provides an interface for OCR using Hidden Markov Models. More... | |
class | cv::text::OCRTesseract |
OCRTesseract class provides an interface with the tesseract-ocr API (v3.02.02) in C++. More... | |
Enumerations | |
enum | { cv::text::OCR_LEVEL_WORD, cv::text::OCR_LEVEL_TEXTLINE } |
enum | cv::text::decoder_mode { cv::text::OCR_DECODER_VITERBI = 0 } |
Functions | |
Ptr< OCRHMMDecoder::ClassifierCallback > | cv::text::loadOCRHMMClassifierCNN (const String &filename) |
Allow to implicitly load the default character classifier when creating an OCRHMMDecoder object. More... | |
Ptr< OCRHMMDecoder::ClassifierCallback > | cv::text::loadOCRHMMClassifierNM (const String &filename) |
Allow to implicitly load the default character classifier when creating an OCRHMMDecoder object. More... | |
Ptr<OCRHMMDecoder::ClassifierCallback> cv::text::loadOCRHMMClassifierCNN | ( | const String & | filename | ) |
Allow to implicitly load the default character classifier when creating an OCRHMMDecoder object.
filename | The XML or YAML file with the classifier model (e.g. OCRBeamSearch_CNN_model_data.xml.gz) |
The CNN default classifier is based in the scene text recognition method proposed by Adam Coates & Andrew NG in [Coates11a]. The character classifier consists in a Single Layer Convolutional Neural Network and a linear classifier. It is applied to the input image in a sliding window fashion, providing a set of recognitions at each window location.
Ptr<OCRHMMDecoder::ClassifierCallback> cv::text::loadOCRHMMClassifierNM | ( | const String & | filename | ) |
Allow to implicitly load the default character classifier when creating an OCRHMMDecoder object.
filename | The XML or YAML file with the classifier model (e.g. OCRHMM_knn_model_data.xml) |
The KNN default classifier is based in the scene text recognition method proposed by Lukás Neumann & Jiri Matas in [Neumann11b]. Basically, the region (contour) in the input image is normalized to a fixed size, while retaining the centroid and aspect ratio, in order to extract a feature vector based on gradient orientations along the chain-code of its perimeter. Then, the region is classified using a KNN model trained with synthetic data of rendered characters with different standard font types.