Image processing, the technology of analyzing images by computer to achieve expected results. Also called image processing. Image processing generally refers to digital image processing. Digital image refers to a large two-dimensional array photographed by industrial cameras, video cameras, scanners and other equipment. The elements of an array are called pixels, and their values are called gray values. Image processing technology generally includes three parts: image compression, enhancement and restoration, matching, description and recognition.
General steps of image algorithm processing
1, image acquisition, the most direct is to sample the measured object through cameras, especially industrial cameras. Here, we also need to design the choice of camera, light source and lens. This is a complicated science, because the quality of the shot image will directly affect the complexity of the subsequent image algorithm, and a good shot image quality will greatly reduce the complexity of the algorithm.
2. After obtaining the image, we need to design an image algorithm, which is divided into many steps:
Image preprocessing, such as image filtering, image enhancement, etc. The purpose of image preprocessing is to remove point noise and avoid some interference factors;
The application of image algorithm, for example: if you are doing edge detection algorithm, you will use some canny operators, for example, if you want to do image segmentation, you will use some image clustering algorithms; For example, to do target detection, at most, mark the work related to network design; For example, to match feature points, we need an algorithm to extract image feature points, such as sift algorithm, so as to achieve our goal.
3, code writing, algorithm design, it is necessary to achieve, including what programming language to choose, such as c++, matlab, python and so on. And what algorithm library to choose, such as opencv, opengl, pillow, etc. ; If you are awesome, you can write your own algorithm without the help of other algorithm libraries, but it is not recommended and time-consuming.
4. After the algorithm is written, it needs to do a lot of testing work, and then test the feasibility of the algorithm. Through continuous debugging, it reaches a stable state, thus realizing a complete algorithm.
5. Write your own algorithm specification or algorithm flow paper. If you want to write a paper, you can add some comparative experiments.