Impression of OpenCV: The function is very powerful, and supports the current advanced image processing technology, the system is very perfect, the operating manual is very detailed, the manual first makes it up the knowledge of computer vision, which covers the mainstream algorithm within the nearly 10 years; then will then Image format and matrix operation, then implement function of each algorithm. I used it to make a Harris angle detector and a Canny edge detector, a total of an hour (first use OpenCV). And the library shows that the image is extremely convenient, two sentences can be. However, the library does not seem to be very stable, and the data data of 32F and 16S, 8U supports BUG. I used the CVFilter2D function to linearly filtering, repeated error, and later checked it was big bug. Later, I used CVMGET to take matrix elements and frequently errors. Carefully check that N-pass ensures that the programs are not problematic, find answers on YahoOGroup: Still bug. . . But I am open, so I can modify themselves; and support CVS. In addition, the library uses the IPL matrix library, and the speed is inexpensive ~~
Impression of CXIMAGE: This development package is fully open source, the image package is a class, the function is extremely powerful, supports excellent in Windows, MFC, supports multiple operations of the image (linear filtering, median filtering, histogram operation, Rotate, regional selection, threshold processing, expansion corrosion, alpha mix, etc.), support from file, memory, or Win32API defined bitmap image format, support to display the image in any window, and the function is very powerful. Moreover, it is very convenient for the operation of the pixel, and there is also a strong DEMO, which can be directly developed, recommended! Disadvantages: There are many subsets inside, which may be more troublesome; and feel slow, it is better to mention the freeImage mentioned later, but the function is really strong!
CIMG: Only one .h file, it is easy to use, but it is better to feel like CxImage. It can be used in conjunction with CXIMAGE because CIMG provides a LAPACK-based matrix calculation function and a perfect linear filter consolidation function, while CIMG is still very convenient to make pixel operations. In addition, unique DISPLAY classes can be easily implemented, including display images, typing, drawing, and more. Also, the library has an optical current-based multi-scale image registration example, very good
FreeImage: C language system, a large number of pointer operations can be guaranteed, including advanced plurality of interpolation algorithms. Also unique to the read of META EXIF information. The biggest feature of the library is to compare, only focus on reading write support for various format images, no display, actually programming, or call API functions to display
