پروژه تشخیص پلاک با متلب
سه شنبه, ۸ بهمن ۱۳۹۲، ۰۱:۰۴ ب.ظ
License Plate Recognition is one state-of-the-art mechanism used nowadays to automate recording information of automobiles. In our AI course we were assigned a project of implementing LPR in Mathworks' MATLAB software.
The project has two phases. In the first phase, we use MATLAB's image processing toolbox to extract plate and then segregate and save digits' images.
In the second phase we are going to use MATLAB's Neural Network Toolbox for recognizing the images. Typical of neural networks, in fact we have a set of training data that is first fed into the system[see the code attached to the article]. Afterwards, the trained network is given new input data by the user and the digit(with the highest probability) is output.
Note that the system I developed, is designed for use inside Iran (persian digits) and only works for digits and not characters. It can further be improved to find alphabetical character too. By the way, this system, like any other system, is not perfect and it seems it doesn't work always. Maybe it needs more training or tweaking the training parameters.
Feel free to modify the code and use it in your projects on condition that you include copyright notice in the code.
Download
حجم: 428 کیلوبایت
password: misaq-tech.blog.ir
The project has two phases. In the first phase, we use MATLAB's image processing toolbox to extract plate and then segregate and save digits' images.
In the second phase we are going to use MATLAB's Neural Network Toolbox for recognizing the images. Typical of neural networks, in fact we have a set of training data that is first fed into the system[see the code attached to the article]. Afterwards, the trained network is given new input data by the user and the digit(with the highest probability) is output.
Note that the system I developed, is designed for use inside Iran (persian digits) and only works for digits and not characters. It can further be improved to find alphabetical character too. By the way, this system, like any other system, is not perfect and it seems it doesn't work always. Maybe it needs more training or tweaking the training parameters.
Feel free to modify the code and use it in your projects on condition that you include copyright notice in the code.
Comments are welcome!
Downloadحجم: 428 کیلوبایت
password: misaq-tech.blog.ir
