4 OCR
Optical character recognition (OCR) generally refers automated alphanumeric character recognition
within image acquisitions comparing the pixel patterns the text areas with familiar learned patterns,
analogous general object recognition (such Matching).6. therefore possible achieve very high recognition rates without
additional training.
• Shape-based Matching finds objects quickly, accurately and robustly. This done creating template reference image and deriving model be
used for teach-in.3 Matching
Matching used locate contours textures images with subpixel accuracy, even they are rotated
or partially covered.
7.
• contrast shape-based Matching, possible find objects with slightly different shapes or
highly textured surfaces, well objects blurred images. Normalized cross-correlation (NCC) used
to grade how well the model matches the image being searched.
Vision function OCR reads and interprets texts the basis several pre-trained fonts suitable for different
applications (dot matrix fonts, SEMI fonts, industrial fonts, handwriting, etc.6.23 41
.
In addition, all segmented characters line are graded terms the quality recognition (Grading-
Value).
Shape-based Matching:
• Shape-based Matching does not use grayscale values; defines the shape the contours instead.Function description
7.
Vision function Matching can performed using one the following Matching methods:
Correlation-based Matching:
• Correlation-based Matching based grayscale values.) well universal font of
characters learned via "deep learning". even works they are rotated,
scaled, perspectively distorted, locally deformed, partially covered, outside the image subject to
nonlinear illumination fluctuations. This method can compensate for both
additive and multiplicative lighting deviations. The model then used search for matches image acquisitions.
Smart Camera Smart Sensor User's manual 1
