Non-standard products

(OCR、OCV…)Equipment and production line application

Identify qualified products to eliminate defective products
Product specification, date, LOGO, product description, qr code, etc.
Character recognition OCR technology development and enterprise application
System equipment detailed description
Optical character recognition, commonly abbreviated as OCR, is a text that can be read by both the machine and the naked eye. Common applications include drug or food packaging date/batch number of tracks, the post office mail sorting and other application document processing, body serial number to read or application in electronics industry and other aspects, and more.
Technical characteristics
The OCR software consists mainly of the following sections.
Image input, pretreatment:
Image input: for different image formats, there are different storage formats, different compression modes. Pretreatment: mainly consists of two values, noise removal and tilting are more
Of camera pictures, most of color image, color image contains huge amount of information, for the content of the picture, we can be divided into foreground and background of simple, in order to make computer faster, better recognition of text, we need to deal with color chart first, make the picture only prospects and the background information, can be simple definition outlook information to black, background information for white, this is where a binarization figure.
Noise removal:
For different documents, our definition of a noise can be different, according to the characteristics of the noise, which is called noise removal
Lean more:
Due to the average user, the photo document, are random, so pictures of pictures come out inevitably produce tilt, this would require the article word recognition software for calibration.
Layout analysis:
The document image segmentation, branch, the process is known as layout analysis, due to the diversity of the actual document, complexity, therefore, didn't have a fixed at present, the most optimal cutting model.
Character cutting:
Because of the limitation of photo conditions often cause conglutination, broken pen, therefore significantly restrict the performance of recognition system, which requires writing knowledge Software don't have character cutting function.
Character recognition:
Earlier with template matching, and later with feature extraction is given priority to, due to the displacement of the text, stroke thickness, broken pen, adhesion, rotation, and the influence of such factors as greatly affect the difficulty of feature extraction.
Page recovery:
People want to identify words after, still lined with the same as the original document images, paragraphs, position unchanged, order the same, the output to a word document, PDF documents, etc., this process is called a page.
Reprocessing and proofreading:
According to the relationship of the specific language context, the recognition result is more positive, which is the post-processing
Application field
Such as a passport processing, security file processing, financial documents, the bill (check), email tracking, publishing, consumer products packaging (single number, batch number, validity) images and clinical application cases

Contact Us

Contacts:Mr Cheng



Address:Software park of xiamen headquarters 3 period 366