Text Recognition
Overview
This text recognition demonstration does the following to get a string representation of text seen through the onboard camera:
- Obtain an RBG image from the camera
- Perform grayscale and binarization threshold processing on the image
- Use the pytesseract text recognition library to recognize the English letters or numbers of the image
- Publish the recognized result to the
/detect_word_reslut
topic
Structure
digraph Text_Recognition_Pipeline { rankdir="LR"; "camera" -> "/rgb_image" -> "detect_node" -> "/detect_word_reslut"; "camera" [shape=ellipse]; "/rgb_image" [shape=polygon,sides=4]; "detect_node" [shape=ellipse] "/detect_word_reslut" [shape=polygon,sides=4]; }Usage
Note
Before running this demo, please make sure that all other programs have been terminated using Ctrl + C.
Start up ROS:
$ roscore
Start the text recognition node from the vision package by entering the command in the terminal:
$ rosrun vision detect_node.py
The output of the text recognition program is published to the topic /detect_word_reslut
. To
inspect the output, enter the following command:
$ rostopic echo /detect_word_reslut