How should Halcon Visual Software learn systematically?

HaLCON is a set of perfect standard machine vision algorithm package developed by German MVtec company. It saves product cost and shortens software development cycle. Halcon: There are many underlying functions and algorithms, and the operation performance is fast. The development needs a certain software foundation and image processing theory. Quick learning methods: research examples, do practical projects. Halcon can not provide the corresponding interface programming requirements. It needs to construct MFC interface with vs or use QT to form a complete set of software. 2. The main requirements are two-dimensional and three-dimensional: (1) recognition and positioning; (2) OCR recognition; (3) two-dimensional code recognition; (4) measurement; (5) defect detection; (6) machine control and other three-dimensional; (1) multi-objective determination; (2) three-dimensional reconstruction; (3) three-dimensional matching; (3) qualified machine vision engineers must have three aspects. Knowledge image processing involves the following fields: A, basic theory of image processing (basic knowledge of image theory), B, image enhancement (contrast stretching, gray transformation, etc.) C, geometric transformation (affine transformation, rotation matrix, etc.) D, frequency domain processing (Fourier transform, DFT, wavelet transform, etc.). High and low pass filter design) E, morphology (expansion, corrosion, open and closed operations, convex hull, etc.) F, image segmentation G, image restoration H, motion image I, image registration (template matching, etc.) J, pattern recognition (classifier training, neural network in-depth learning, etc. (2) Software programming has C, C++, C# and M. The foundation of FC interface development A and C language learning mainly depends on C language related knowledge B and C++ written by Tan Haoqiang. C++ mainly focuses on C++ primerplus books C and MFC learning. It mainly focuses on Sun Xin's "VC++ in-depth interpretation" book and corresponding video tutorials. It also writes programs and practices on VC++ 6.0 software or VS2010 software. D, C # can read books "C # from Introduction to Proficiency", "C # Introduction Classics", "C # Graphics Course" (3) Optical knowledge: Optical knowledge plays a key role in the selection of cameras, light sources, lenses, etc. when you design a project. 4. The most important way to learn HALCON is to learn the methods, processes and routines of HALCON. The main flow routines of image algorithm are as follows: feature extraction summary: 1. geometric features (area, circumference, rectangularity) 2, texture features (correlated with gray level, such as entropy, energy value) 3, color features 4, probability features 5, operator description. Describe feature 6. Hough feature (gradient histogram feature) for machine vision project. When we get the project, we usually follow the following process: first step: requirement analysis; second step: algorithmic process planning and business logic design; third step: modular programming and integration; fourth step: debugging, according to feedback. As a result, the program Bug is constantly modified to meet customer needs, and finally the software and hardware operation documents are delivered to customers. A good way to learn machine vision: 1. Learning machine vision must be combined with the actual project, learn and summarize the experience and lessons in practice, and systematically learn the required knowledge. 2. To supplement some knowledge of C++ and C#, to develop VS jointly, to study and debug on site by customers, to constantly study and analyze examples, and to master the routine process of methods. 3. According to practical problems, learning module is called and learning is done according to the method routine. 4. It's better to use HALCON to realize the image processing part first, and then use MFC graphical interface to realize it in VS2010 development software. Practical learning is the best way. The development trend of artificial intelligence machine vision is to combine neural network and deep learning. CV Vision Network and Halcon Learning Network have some information. Finish the summary.