OpenCV learning code by Python
code: Readcode.py
In dBarcode recognition: It is easy to do. I used cv2.getStructuringElement to check out the rectangle as a kernel, and remove another noise. However, the QRcode is another case before I was wrong that I thought it's as easy as dBarcode. The QRcode,we need conform the THREE small rectangle frist. The way is 5 more times iteration of hierarchy.
(2018/3/8,USA)
Reference: http://dsynflo.blogspot.in/2014/10/opencv-qr-code-detection-and-extraction.html
deconvolution is an iterative algorithm for recovering an image which is blurred by a known point spread function (PSF). You can find the iterative algorithm steps in：https://en.wikipedia.org/wiki/Deconvolution https://en.wikipedia.org/wiki/Richardson%E2%80%93Lucy_deconvolution
作用：把模糊画面还原， 在移动设备中，提高识别成功率：
关于motion blur 的论文： http://www.cse.cuhk.edu.hk/%7Eleojia/projects/robust_deblur/index.html
Wiener_deconvolution： https://en.wikipedia.org/wiki/Wiener_deconvolution
opencv_4.py
deconvolution.ipynb
BGR->HSV Example: Tracking the Blue pen
1 Collect "Negative" or "background" image
2 collect or create "positive" images
3 Create a positive vector file by stitching together all positives.
opencv_createsamples -img XXX.jpg -bg bg.txt -info .ist -pngoutput info -maxxangle 0.5 -maxyangle -0.5 -num -1950
4 Train cascade
opencv_traincascade -data data -vec positives.vec -bg bg.txt -numP os 1800 -numNeg 900 -numStages 10 -w 20 -h 20