Text- Recognition from images in Design & Analysis Algorithm
- 1. Topic: Text-Recognition From Images: PRESENTED BY: SADIA ZAFAR 170403
- 2. Code Implementation: Classes Used: Camera Source, Surface View, Text Recognizer, TextToSpeech and Detector
- 3. Working: First the CameraSource gives the camera view on the screen. Then CameraView combined with the CameraSource sets the properties of camera screen for example, Back facing camera or front facing camera, frames per second, Autofocus should be enabled or not and dimensions. CameraSource class also sends Frames per second to the Detector class synchronously in a pipeline Structure and it checks if the Detector object can not process given frames per second then it reduces the frames per second because it is synced with the Detector. Now In the TextRecognizer API, Detector class from the help of Detector.
- 4. Processor detects these frames and recognizes text with the help of algorithm explained above. This text is detected as TextBlocks and saved in a SparseArray which using StringBuilder class makes a string which represents the whole text in an image. Now this text is given to TextToSpeech which speaks the text to the user. Two Frames per second are being read by the Detector on live camera and for each frame a thread is being run to detect text in the frames(Images) taken from the camera.
- 5. Thank you