Let a professional practitioner answer this question. Just happened to be myself.
If from the overall planning of life should consider the following questions:
1 The reality is forced, you are ready to develop in which city (to buy a house, girlfriend's work, etc.), the city whether there are sufficient image processing industry to choose. For example, Beijing, Shanghai and Shenzhen, Beijing's image companies feel like nearly half of the country. Years of recruiting feel. The rest of Guangzhou, Wuhan, Chengdu also has a few image companies. Like the city where I am, there are more universities, more image learners, no industry, no research institute engaged in the direction of the image. You're a Ph. The school's pit you are the vast majority is not able to enter.
2 Study image of graduate students can engage in the specialty of less than 15%, this is my side of the statistics. Image analysis is more affected by the environment, such as lighting. This is a different story, it is more difficult and less reliable. So it has a long way to go as a detection tool. If you don't intern at a company that does image and gain trust then your chances of getting employed with image processing go way down. How many practical image algorithms are there in the world every year? Those of you who send out papers know the constraints of your algorithms. The state of the art paper, how many scenarios can be applied. So don't be superstitious about how awesome your algorithm is. Improve your programming skills, take a look at graphics stuff (don't ask me what's the difference between an image and a graph.), play with parallel computing, play with the graphics. I'm not sure if I'm going to be able to do that, but I'm going to be able to do that, and I'm going to be able to do that.
3 To the subjective place, talk about the positive. I'm not sure if you're going to be able to do that, but I'm sure you'll be able to. I'm not sure if you're going to be able to do that, but I'm sure you'll be able to. But are you qualified for the position it needs. Back to the image to say, human access to information 80% of the information is obtained through the image, you say the image is important. Image processing, pattern recognition direction is indeed relative to the communications electronics is narrow, but you say that the employment of scientists to build the atomic bomb is not narrow. Believe in your own specialty, the longer the more irreplaceable, you can take the route of technical experts, more free. Automation, artificial intelligence is the general direction of future development, image is an important means. Work experience tells me that the image processing this thing is not in the number of people, the value of the depth of the essence. Don't do it alone, a small and precise team combat power is very strong, but just need one or two people to do image algorithm company is not recommended to go. Weigh whether you have the enthusiasm, brains, math skills, and communication skills. If you have entered the relevant enterprises, program-related things can be learned to train (see the company to have the foresight and courage of course you have to have the brains and math base), analyze the problem, problem-solving ability is very important, and more importantly, to raise the issue. Students who study well are better at the first two, and the image of this innovative work is more important to raise questions, this is the performance of innovative thinking. This is not a slogan. Of course, being able to ask questions depends, on the depth and breadth of your study of the problem, and ultimately on your interest in the study.
Finally, a few words about the disciplines of image processing research.
1 Medical 2 Recognition Classes 3 Part Inspection 4 Satellite Imagery.
In general the medical population has a high employment volume and relatively more companies. For other images, medical images are better standardized with less impact on the outside world and less disparity between different devices. After all, they are the credentials of diagnosis. So if you want to jump ship (hehehe), the same category algorithms basically do not need to delete are applicable (pay attention to intellectual property rights oh).
Down to the recognition category, all the text, pedestrians, vehicles and so on need to recognize the tracking object. Do a good job of making a fortune, such as a U.S. company's fingerprint detection after 9/11, the speed and accuracy of the best. It got a big government order, and it paid off handsomely.
But it's not easy to do well. However, this is my favorite, the most artificial intelligence feeling. This is the highest form of automation class professional research image.
3 parts detection, including some food packaging type of foreign object detection. For industrial automation is also a very promising direction, some are even used for chip-level detection, if the microscopic image is more involved in sub-pixel and three-dimensional correlation.
4 satellite images. Generally the country needs, if possible, some research institutes have related categories. The company, not too clear
On the other hand, I forgot to mention the more important one. English. This does not bother to say. Chinese paper can be a popular science, want to do algorithmic research, you can not read is obviously not allowed. If the overall strength of English NB, you can go to some foreign enterprises. Objectively speaking, Europe and the United States enterprises are still good drops, attach importance to people, create, do algorithms have patience. The salary is also relatively generous. This kind of enterprise seems to be more Shanghai side
In summary, if you do not have the patience to see the end, go to find the image of the company's internship, the premise is that the hands of some of the program to do the work, this is really not difficult. , play with opencv, and then feel it out for yourself. There is also the accumulation of contacts. This is also extremely important. Just out of school without the benefit of a certain degree, and no work experience, the importance of internship experience is beyond imagination.
A word of advice, if you have no interest in images or do not do it. You can't do it well either.
Add one more sentence, I have now done medical class, recognition class, done a little bit of detection class. I've also been hired, and I've been hired by others. Give you a reference.