Introduction to machine vision software evision

Introduction to machine vision software evision

Evision is a machine vision Software Development SDK launched by euresys, Belgium. Compared with other Machine Vision Development kits such as cognex
Vision library, matrox Imaging
Library, which seems to provide more options than the two in the SDK function classification, and this feeling matrox Imaging
Library is very poor and pale.

Evision machine vision software development kit allCodeAll of them are optimized by MMX commands, and the processing speed is very fast. I feel like it is better than Intel's IPP (of course, it is better than IPP,
After all, this is developed by Intel), but it provides many more machine vision functions than IPP, such as OCR, OCV, Image Quality Detection Based on Image Comparison, barcode and
Matrixcode recognition. In addition, I have developed machine vision software based on evision in the past two years. through development, I found that evision is based on OCR and geometric shapes.
The configuration above is relatively weak, and the strongest is the template matching based on gray correlation and image quality detection based on image comparison. In addition, evision is easier to use than cvl and mil.
Function, among so many development kits, evision is the easiest to use. For example, for a machine vision project that detects character defects, the evision code segment is only about 50 lines, while
At least 100 lines are required for cvl, and the functions are basically similar (but cvl is a little better, but evision provides us with enough support for Machine Vision Development of more than 95%.
Fields ).

Evision consists of the following modules:

Easy-general library, including definition of data structures and storage and reading of image files.

Easyimage-General Library for digital image processing, including general digital image processing operations andAlgorithmSuch as FFT, Image Algebra, histogram statistics and analysis, image registration, and geometric transformation.

Easycolor-Color Image Processing library, including color image space conversion, Bayer conversion, and K-means-based Color Image Segmentation Algorithms

Easyobject-blob analysis library, including blob Feature Extraction and image segmentation.

Easygauge-image measurement tool based on sub-object.

Easymatch-the image matching package based on gray correlation is fast, and the matching result of sub-object precision can be obtained. For rotation, scaling, and shifting, the position of the template image can be precisely found.

Easyfind-image matching packages based on geometric shapes are fast, but the accuracy is not accurate. Due to the impact of image noise, euresys is still improving.

Easyocr-the Character Recognition toolkit is not very good either, because it is based on the template matching method and is not precise based on neural networks, but it is very suitable in most cases, fast, high positioning accuracy.

Easyocv-the Printing Quality Inspection package is particularly well done. It can be widely used in printing and character quality testing scenarios with high speed and accuracy.

Echecker-wider range of printing quality inspection packages, suitable for all printed inspection objects.

Easybarcode and easymatrixcode-one-dimensional, two-dimensional Barcode recognition library. It is also very first-class, comparable to cvl.

Easybga-bga detection package for semiconductor chips, which has never been used.

Easyworldshape-Computer Vision Calibration tool. Alas, if I had been familiar with this stuff earlier, my graduation thesis would have been much easier to write.

After reading the above introduction, do you have a heart feeling? I have been shocked n times (n> = 10)