How to learn machine vision and what are the machine vision algorithms

In the process of communicating with everyone, I found a problem and I have always wanted to say it. But usually the truth is not so pleasant, so there is always some hesitation. But after thinking about it, since the purpose of everyone is just to discuss problems and exchange ideas, it should not hurt to say it.

The problem is: I found that 80% of my friends who are engaged in this industry (machine vision) in China are taking the wrong path.

This feeling comes from the questions that people ask on the forum. Because. It's quite ridiculous to see a lot of questions asked by friends in the forum.

Ridiculous does not mean that the questions asked are too simple and naïve, but that the questions asked look copied from a book and are not encountered in the actual situation. In other words, most of the people in the forum are just talking on paper, and very few people are actually developing their own vision systems. I say this because people who are developers themselves will never be able to ask some strange questions.

To go a little deeper, it seems that friends in China only like to nibble on books, and are not willing (or may be too lazy) to do it.

The development of machine vision at home and abroad is different. In my own opinion, only by first understanding the differences between the two sides can we explain how to start learning. The development of foreign machine vision to this day has developed from a "one-package-to-the-end" work procedure to a stage of detailed division of labor.

Due to space issues. I will not talk about how this industry has developed from "one package to the end" to a detailed division of labor.

In a word, the development of foreign machine vision to this day can be clearly divided into three parts:

1. The underlying development part.

2. Secondary development part.

3. End-use part.

So abroad, people who are engaged in this industry can now be divided into three types of people simply and clearly:

1. People who work on the underlying development (people who work on the underlying development). 2. People who are engaged in secondary development (people who are engaged in secondary work).

3. People who use and operate machine vision systems (people who work on end-use work).

Machine vision. basic steps of the algorithm;

1. Image data decoding.

2. Image feature extraction.

3. Identify the target in the image.

Machine vision is a branch of artificial intelligence that is developing rapidly.

To put it simply, machine vision is the use of machines instead of human eyes to make measurements and judgments.

The machine vision system converts the acquired target into an image signal through a machine vision product (i.e., an image acquisition device, divided into CMOS and CCD), transmits it to a dedicated image processing system, obtains the morphological information of the photographed target, and converts it into a digital signal according to the pixel distribution, brightness, color and other information; The image system performs various calculations on these signals to extract the characteristics of the target, and then controls the action of the equipment in the field according to the results of the discrimination.

Now there are many companies doing visual inspection, both at home and abroad, and many of them are very good.

It can provide complete machine vision software solutions, and can also provide customers with algorithm-level customization, covering all industrial application fields, with a wide range of applications. The application of machine vision will become more and more because the level of computing is getting higher and higher, and more complex vision algorithms can be processed; In fact, many things, including the now popular GPS, were first done by foreign companies, and the procedures were outsourced by Chinese;

Personally, I think the application of opto-electromechanical is very mature, and there will be no new things.

Image Preprocessing Algorithms:

1.Image gradient algorithms Sobel, Scharr

2.Corner detection algorithm Harris

3.Edge detection algorithm canny

Straight line detection algorithm hough

Image Feature Processing Algorithms:

scale-invariant feature transformation).

Acceleration robust features).

gradient histogram).

histogram) local binary mode).

6.Brute-Force (brute-force feature matching) common algorithms for machine learning:

Nearest neighbors) support vector machines).

mean) cascading classifier.

Random Forest) 6decision tree logistic regression).

8.gmm (Gaussian Mixed Model).

Self-organizing mapping).

matching matrix) convolutional neural network).

Combination of theory and practice.

The most basic feature of a machine vision system is to increase the flexibility and automation of production. In some dangerous working environments that are not suitable for manual work or where artificial vision is difficult to meet the requirements, machine vision is often used to replace artificial vision. At the same time, in the process of large-scale repetitive industrial production, the use of machine vision inspection methods can greatly improve the efficiency and automation of production.

In 2011, China's machine vision market entered a period of post-growth adjustment. Compared with the rapid growth in 2010, although the growth rate has decreased, it still maintains a high level. In 2011, the size of China's machine vision market was 100 million yuan, a year-on-year increase, and the growth rate decreased by one percentage point compared with 2010, of which smart cameras, industrial cameras, software and boards maintained a growth rate of not less than 30%, and the light source also reached a growth rate, which was much higher than the growth rate of China's overall automation market.

The electronics manufacturing industry continues to be the main driver of rapid demand growth. In 2011, the market size of the machine vision product electronics manufacturing industry was 100 million yuan, growing. The city share was reached.

Electronics manufacturing, automotive, pharmaceutical, and packaging machinery account for nearly 70% of the machine vision market share.

A typical industrial machine vision system includes: light source, lens (fixed focal length lens, zoom lens, telecentric lens, microscope lens), camera (including CCD camera and COMS camera), image processing unit (or image capture card), image processing software, monitor, communication input and output unit, etc.