PATTERN EXPERT Image Processing Engine (IPE)

Image Import / Image Export

The IPE is able to import raster as well as vector graphics from files. The currently supported input file formats are: BMP, DXF, GIF, JPG, PCD, PCX, PNG, and TIF. Images can also be scanned via the Twain interface. The loaded images can be organized in image series for comfortable batch processing.
Currently, the processed images can be saved using the following output file formats: BMP, GIF, JPG, PCX, PNG, SVG, and TIFF.

Image Acquisition

The IPE supports a number of framegrabbers as well as the FireWire, IEEE 1394, interface so that images can be directly imported from a camera.

Image Transformations

A variety of image operators are available for image processing. Among these are the classical matrix operators like Sobel and Laplace filters for improving focus and edge detection. Particular combinations of matrix operators yield powerful tools such as shading correction.
Another group is formed by the morphological operators which are important for a number of applications like e.g. skeletonizing of characters, among others.
Finally, the Fast Fourier Transformation (FFT) should be mentioned here.

Macros

Fixed sequences of consecutive image operations can be defined as macros. Once a macro is constructed using the IPE macro editing tool, it can be executed on either one image, or on a series of images (batch mode).

Image Analysis

In image analysis a number of specific quantities or features is calculated from the image. These may include the sizes of objects, their areas, their maximum diameters, or other quantities. The features may as well contain values derived from the FFT transform, which describe the global structure of an image.

Object Recognition and Object Classification

Object recognition and classification is done according to so-called "Classification Schemes". A classification scheme contains information about which objects or image sectors are to be classified, which classes they are to be assigned to (e.g. by color or by shape) and which classifier algorithm is to be used - e.g. a probabilistic classifier, an Artificial Neural Network, or a Support Vector Machine.

Machine Vision - Industrial Image Processing

Another integral part of the IPE is a framework for MachineVision applications. These are computer-aided systems which are mainly used in quality control. Here, the system uses a camera to take an image of the object to be inspected, it analyses the image using appropriate methods of image processing, and communicates the result to some other system - the complete process is fully automatic.
The IPE supports external devices (e.g. photoelectric barriers) and contains a framework for flexible and configurable output of results via network (LAN) or via the serial port.
The IPE's MachineVision framework is used in our reader system (for Data Matrix / Dot Matrix and PDF417) PATTERN EXPERT legeo.

Optical Character Recognition - OCR

Based on the IPE, PATTERN EXPERT has developed a powerful framework for Optical Character Recognition. This OCR engine forms the heart of the software SENSDOC®.