News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
Image recognition project
Image recognition project
License plate number recognition system:
Recognition steps: license plate location, character segmentation, character recognition.
The method of license plate location, based on the method of the line edge detection, the method based on the threshold iteration, the method based on the color information and the method based on the gray level detection, based on the neural network.
Character recognition: feature extraction and template matching.
The whole recognition process includes image pre-processing and license plate recognition.
Image preprocessing includes two values, denoising, license plate location, character segmentation and character refinement. The two value is actually converting the gray image to 0-1 (255) pixel image. De-noising is to remove some noises and identify easily, and use the out of limit neighborhood mean method to de noise the two valued. License plate location is the license plate extraction. Here, we use transverse and vertical scanning to extract the license plate. The standardization of the license plate image before the character segmentation. Character segmentation: divide the characters one by one, and distinguish the two by using the characteristic that the spacing between words is usually larger than the interval of self discouraging. Character refinement: the purpose is to get the image which is similar to the original area shape and is composed of simple arcs and curves.
License plate recognition: the commonly used methods are based on template matching and neural network based method. Template matching has strong fault tolerance and is suitable for strong interference, but it is slow to recognize the speed, and it is difficult to meet the requirements of real-time. The method based on neural network has a faster recognition speed, especially for two valued graphs, which is faster and can meet real-time requirements.
One dimensional bar code recognition:
One dimensional bar code is a mark made up of different rules, such as different sizes, black and white intersections, numbers and characters.
Usually one dimensional bar code expresses information in only one direction, generally the horizontal direction. For each item, the encoding is unique. Several one - dimensional bar code code with the highest frequency at present has EAN, UPC, 39 code, cross 25 code and EAN128. The code standard used in our country is EAN code.
There are two versions of the EAN barcode. One is 13 bit standard barcode, the other is 8 bit shortening bar code. EAN-13 barcode is composed of barcode symbols that represent 13 digit codes. 13 of the barcodes are processed here.
The EAN barcode is made up of the left blank area, the start symbol, the left data symbol, the middle separator, the right data character, the check character, the terminator and the right blank area. Its structure is shown in the following figure:
Recognition steps: license plate location, character segmentation, character recognition.
The method of license plate location, based on the method of the line edge detection, the method based on the threshold iteration, the method based on the color information and the method based on the gray level detection, based on the neural network.
Character recognition: feature extraction and template matching.
The whole recognition process includes image pre-processing and license plate recognition.
Image preprocessing includes two values, denoising, license plate location, character segmentation and character refinement. The two value is actually converting the gray image to 0-1 (255) pixel image. De-noising is to remove some noises and identify easily, and use the out of limit neighborhood mean method to de noise the two valued. License plate location is the license plate extraction. Here, we use transverse and vertical scanning to extract the license plate. The standardization of the license plate image before the character segmentation. Character segmentation: divide the characters one by one, and distinguish the two by using the characteristic that the spacing between words is usually larger than the interval of self discouraging. Character refinement: the purpose is to get the image which is similar to the original area shape and is composed of simple arcs and curves.
License plate recognition: the commonly used methods are based on template matching and neural network based method. Template matching has strong fault tolerance and is suitable for strong interference, but it is slow to recognize the speed, and it is difficult to meet the requirements of real-time. The method based on neural network has a faster recognition speed, especially for two valued graphs, which is faster and can meet real-time requirements.
One dimensional bar code recognition:
One dimensional bar code is a mark made up of different rules, such as different sizes, black and white intersections, numbers and characters.
Usually one dimensional bar code expresses information in only one direction, generally the horizontal direction. For each item, the encoding is unique. Several one - dimensional bar code code with the highest frequency at present has EAN, UPC, 39 code, cross 25 code and EAN128. The code standard used in our country is EAN code.
There are two versions of the EAN barcode. One is 13 bit standard barcode, the other is 8 bit shortening bar code. EAN-13 barcode is composed of barcode symbols that represent 13 digit codes. 13 of the barcodes are processed here.
The EAN barcode is made up of the left blank area, the start symbol, the left data symbol, the middle separator, the right data character, the check character, the terminator and the right blank area. Its structure is shown in the following figure: