The images to be processed may come from a frame grabber, a scanner, a file, or IEEE1394 (Firewire), GigE Vision or USB cameras. Color and monochrome images are supported.

Open eVision 1.2 comes with an alternate set of C++ headers and an ActiveX component that allows developing or porting code against the older API that was supplied with eVision 6.7.1 (and lower) and Open eVision 1.0. See the "Migration Guide" pdf document for detailed information.

Accuracy is of prime importance to most machine vision applications. To meet this requirement, Open eVision’s metrology (EasyGauge), blob analysis (EasyObject), standard pattern matching (EasyMatch) and geometric pattern matching (EasyFind) functions all return results with sub-pixel accuracy.

To further increase the precision of the measurements, Open eVision includes an extensive set of calibration functions. These functions, after a calibration phase that can be performed from the image of a precise dot grid or from a list of point coordinates, compensate all measurements for lens distortion (correction of pin cushioning, barreling), non-square pixels (correction of the pixel aspect ratio) and rotation and perspective distortions.

They also provide the capability of working in real-world measurement units (such as mils, microns or millimeters), and support the conversion of coordinates from pixel coordinates to real-world coordinates and vice-versa.

Open eVision supports loading and saving images from/to the disk. In addition to the existing image formats -BMP, JPEG, TIFF- Open eVision now supports JPEG-2000, PNG (a supports lossless compression) and Serialized (proprietary image file format obtained from the serialization of the Open eVision image objects) formats.

On top of this, Open eVision includes an automatic conversion of the file format regardless of the pixel format from an image on the disc into the best matching Open eVision image object type.

The processing speed of an image can be accelerated by focusing on a specific region of the image (Region of Interest) avoiding interferences from the remainder of the image. The number of pixels to consider is then reduced.

The processing of all Open eVision functions can be restricted to a Region of Interest (ROI). Open eVision supports nested rectangular ROIs, which are organized in a hierarchical way in each image.

To add flexibility to the shape of the ROI, Open eVision supports Flexible Masks for selected functions of the EasyObject and EasyImage libraries. A mask represents a two-class segmentation of pixels which separates the associated image in do-care areas (that must be considered) and don’t-care areas (that should not be considered). Flexible masks support complex and disconnected shapes.

Open eVision also includes functions to help display the result of the image analysis, and provide interactivity with the user. For example, Regions Of Interest, EasyGauge measurement gauges, blobs computed with EasyObject, patterns located with EasyMatch and EasyFind, characters and codes read with EasyOCR, Easy- MatrixCode and EasyBarCode can easily be displayed on top of the image.

Open eVision supports modern error tracing functions, and provides precise execution time measurement functions for application profiling.

Open eVision is multi-thread safe. This means that it is designed to support simultaneous execution by multiple threads on the same CPU. This is particularly suitable if your application includes independent tasks and allows them to be executed simultaneously. But this also requires that your application accepts sharing piece of data to be accessed by only one thread at any given time. Then, you can design such applications so that each part is controlled by a separate thread. As many threads as required can be created.

Open eVision are 64-bit and 32-bit libraries that require a processor compatible with the x86 instruction set, with MMX extensions. If the SSE or SSE2 extensions are present, they are used, but they are not required.