This question should be answered by professionals. I happen to be. . But the analysis is still too narrow.
If we look at the overall planning of life, we should consider the following questions:
Due to the reality of 1, which city are you going to develop (buying a house, working for your girlfriend, etc.)? ), and whether the city has enough image processing industries to choose from? For example, in Beijing, Shanghai and Shenzhen, the image enterprises in Beijing account for nearly half of the country. The feeling of years of recruitment. The rest of Guangzhou, Wuhan and Chengdu also have a few image enterprises. Like the city where I live, there are many awesome universities, and it is useless for me to study more images. There is no industry or research institution engaged in image direction. You are a doctor. . Can't get into the school pit.
2 Less than 15% of graduate students majoring in imaging can engage in this major. This is the statistics around me. Image analysis is greatly influenced by the environment, such as illumination. To put it another way, it is more difficult and unreliable. So there is still a long way to go as a means of detection. If you don't go to an internship in an image company and gain trust, your chances of getting an image processing job will be greatly reduced. How many practical image algorithms can there be in the world every year? Comrades who send papers, you know the binding nature of your own algorithms. It is the most advanced paper, and there are many suitable scenes. So don't be superstitious about how awesome your algorithm is. Improve your programming level and have a look at the graphics (don't ask me about images, what's the difference between graphics). . ), play parallel computing, embedded, and expand employment.
3 Go to the subjective place and talk about the positive. Is the company short of people? Every company lacks it. But whether you are qualified for the position it needs. Back to images, 80% of the information obtained by human beings is obtained through images. Do you think image is important? The direction of image processing and pattern recognition is indeed narrower than that of communication electronics, but do you think the employment of scientists who build atomic bombs is narrow? . The longer you believe in your major, the more irreplaceable it becomes. You can take the route of technical experts and be more free. Automation and artificial intelligence are the general direction of future development, and images are an important means. Work experience tells me that image processing is not crowded, but profound. Don't go it alone. A small and sophisticated team is very powerful, but it is not recommended for companies that only need one or two people to do image algorithms. Consider whether you have enthusiasm, brains, math skills and communication skills. If you have entered the relevant enterprises, you can learn and cultivate things related to procedures (depending on whether the company you are applying for has the foresight and courage, of course, you must have a brain and mathematical foundation). The ability to analyze and solve problems is very important, and more importantly, the ability to ask questions. Students who study well do better in the first two items, while the image innovation pays more attention to asking questions, which is the performance of innovative thinking. This is not a slogan. Of course, whether you can ask questions depends on the depth and breadth of your research, and ultimately depends on your interest in research.
Finally, talk about the category of image processing research.
1 medical 2 identification 3 parts inspection 4 satellite images.
Generally speaking, the amount of medical workers is relatively large, and there are relatively many enterprises. For other images, the standardization of medical images has a better impact on the outside world, and the gap between different devices is relatively small. After all, it is a proof of diagnosis. So if you want to change jobs (hehe), the algorithm of the same category is basically applicable, and you don't need to delete it (pay attention to intellectual property rights).
Come down to identify the class, all people, pedestrians, vehicles and other objects that need to be identified and tracked. Make a fortune. For example, the fingerprint detection of an American company has the best speed and accuracy after 9 1 1. I got a large government bill, which paid off handsomely.
But it's easier said than done. . However, this is my favorite and the most artificial intelligence feeling. This is the highest form of image research for automation majors.
3. Inspection of parts, including inspection of some foreign bodies in food packaging. It is also a promising direction of industrial automation, and some of them are even used for chip-level detection. If microscopic images are involved, most of them involve sub-pixel and three-dimensional correlation.
4 pictures of Zhang Weixing. If possible, the general country needs to study all relevant categories. The company's words are not clear.
By the way, I forgot to mention a more important one. English is needless to say. China paper can popularize science. If you want to do algorithm research, it is obviously impossible for you to understand. If the overall strength of English is NB, you can go to some foreign companies. Objectively speaking, European and American enterprises are still good, paying attention to people and patiently creating and doing algorithms. The salary is also relatively generous. There seem to be more enterprises like Shanghai.
To sum up, if you don't have the patience to finish reading it, go to the imaging company for an internship. This premise is that you have some skills in making programs, which is really not difficult. . Play opencv and feel for yourself. And network. This is also extremely important. Just out of school, to some extent, not interested, no work experience. The importance of internship is beyond imagination.
A word of advice, if you are not interested in images, don't do them. You can't do it well either.
I still have a say in your specific 3D problem. If you get good grades, you can't say how many companies want you in employment, at least you will never worry about your job. Of course, you should be willing to recruit the city where your company is located. Learn the transfer function of rendering, learn parallel operations such as cuda, and learn opengl. Not much money: 10k/ month. . But you didn't graduate that much. For a first-class city, let alone the monthly salary.