Image Character Recognition Using Signal & Pattern Analysis

Download Document/Synopsis
Optical character recognition is conversion of images of typed, handwritten or printed text into machine-encoded text from image. Here we come up with system where pattern of the image is extracted and image text is recognized using Optical character recognition in matlab. System will recognize the pattern and will detect the characters in image. And will display those characters as output. Since image can be scanned photos or other images. We applied image preprocessing steps in order to extract only text from image. Image patterns will be stored in another directory. The query image is compared with image patterns stored in database. If image matches with the pattern stored in directory, system will apply algorithm to detect the text and display those text as an output. User will input the image. Since images are more susceptible to noise and other environmental interference so we first have to remove noise by applying filter. Image pre-processing steps are applied on images. Unwanted objects are removed. Bounding boxes is applied to text extracted. These text are in image format. These images are converted to characters. System uses optical character recognition to extract characters from image. Character and number images are stored in directory. The extracted text image are separated by bounding box. Each bounding box will contain each character or number. Each character or number is resized to image stored in directory. Extracted image and existing character image feature is compared. After comparison characters are detected. Finally detected characters are shown in text format.


Image Character Recognition


Advantages
  • OCR method detects text more accurately.
  • Used many image pre-processing steps to remove noise and unwanted objects.
Disadvantages
  • Problem while detecting characters with similar features
-->