1、import cv2 as cvimport numpy as npfont = cv.FONT_HERSHEY_SIMPLEXimage = cv.imread("c:\\book.png")cv.imshow('image', image)加载图片
2、retval, thresholdImagre = cv.threshold(image, 100, 255, cv.THRESH_BINARY)imgfont = cv.putText(thresholdImagre, 'thresholdImagre: 100,255', (100, 100), font, 1.2, (255, 255, 255), 2)cv.imshow('thresholdImagre', thresholdImagre)对彩色源图进行100-255阈值化
3、gray = c箪滹埘麽v.cvtColor(image, cv.COLOR_BGR2GRAY)retval1, threshold = cv.thr髫潋啜缅eshold(gray, 100, 255, cv.THRESH_BINARY)imgfont = cv.putText(threshold, 'threshold: 100,255', (100, 100), font, 1.2, (255, 255, 255), 2)cv.imshow('threshold', threshold)转化成灰度后二值化处理和上图比较：图片都变清晰但是边缘存在彩色阴影或者黑白阴影
4、adaptiveThresholdMean = cv.adaptiveThreshold(gray,255,cv.ADAPTIVE_THRESH_MEAN_C,cv.THRESH_BINARY,13,9)imgfont = cv.putText(adaptiveThresholdMean, 'adaptiveThresholdMean: 13,9', (100, 100), font, 1.2, (255, 255, 255), 2)cv.imshow('adaptiveThresholdMean', adaptiveThresholdMean)采用自适应平均大法。二值化，窗口大小选13，常数c选择9
5、adaptiveThresholdGaua = cv.adaptiveThreshold(gray, 255, cv.ADAPTIVE_THRESH_GAUSSIAN_C, cv.THRESH_BINARY, 13, 9)imgfont = cv.putText(adaptiveThresholdGaua, 'adaptiveThresholdGaua: 13,9', (100, 100), font, 1.2, (255, 255, 255), 2)cv.imshow('adaptiveThresholdGaua', adaptiveThresholdGaua)采用自适应高斯大法。二值化，窗口大小选13，常数c选择9
6、小结：采用adaptiveThreshold函数能够明显改善清晰度，并且边角也比较清晰。注意：但是窗口和常数的选择会影响效果。比如 窗口和常数 13， 1 效果马上变坏。
