008 对比度调整-直方图均衡
本文共 2325 字，大约阅读时间需要 7 分钟。
介绍
YUV：Y分量确定颜色的亮度（称为亮度或亮度），而U和V分量确定颜色本身（色度）
YUV分通道显示 import cv2 as cvimport matplotlib.pyplot as pltimg = cv.imread('../images/scene_001.jpg')yuv = cv.cvtColor(img, cv.COLOR_BGR2YUV)y_img = yuv.copy()y_const = 255y_img[:, :, 1] = y_consty_img[:, :, 2] = y_constu_img = yuv.copy()u_img[:, :, 0] = y_constu_img[:, :, 2] = y_constv_img = yuv.copy()v_img[:, :, 0] = y_constv_img[:, :, 1] = y_constfig, ax = plt.subplots(2, 2, figsize=(8, 8))ax[0][0].set_title('origin')ax[0][0].imshow(cv.cvtColor(yuv, cv.COLOR_YUV2RGB))ax[0][1].set_title('y channel')ax[0][1].imshow(y_img[:, :, 0], cmap='gray')ax[1][0].set_title('u channel(y=' + str(y_const) + ')')ax[1][0].imshow(cv.cvtColor(u_img, cv.COLOR_YUV2RGB))ax[1][1].set_title('v channel(y=' + str(y_const) + ')')ax[1][1].imshow(cv.cvtColor(v_img, cv.COLOR_YUV2RGB))[axi.axis('off') for axi in ax.ravel()]plt.show() 
将RGB转成YUV，直方图均衡化Y，重新合成图片 代码实现（Python）
import cv2 as cvimport numpy as npimport matplotlib.pyplot as pltimg = cv.imread('../images/scene_001.jpg')yuv = cv.cvtColor(img, cv.COLOR_BGR2YUV)hist_size = 256"""calcHist(images, channels, mask, histSize, ranges[, hist[, accumulate]]) -> hist"""y_hist = np.array(cv.calcHist(img, [0], None, [hist_size], [0, hist_size - 1])).reshape(hist_size)fig, ax = plt.subplots(5, 2, figsize=(8, 8))ax[0][0].set_title('origin')ax[0][0].imshow(cv.cvtColor(img, cv.COLOR_BGR2RGB))ax[0][0].axis('off')ax[0][1].set_title('y hist')ax[0][1].hist(y_hist, 50)# Equalize histdef equalize(adjust_ratio, ii): rows, cols, _ = img.shape y_hist_equal = y_hist total = 0 for i in range(len(y_hist)): total += y_hist[i] tmp = total * 255 / rows / cols y_hist_equal[i] = int(min(max((tmp - i) * adjust_ratio + i, 0), 255)) yuv_equal = yuv.copy() for row in yuv_equal: for col in row: col[0] = y_hist_equal[int(col[0])] y_hist_equal = np.array(cv.calcHist(yuv_equal, [0], None, [hist_size], [0, hist_size - 1])).reshape(hist_size) ax[ii][0].set_title('Equalize(ratio=' + str(adjust_ratio) + ')') ax[ii][0].imshow(cv.cvtColor(yuv_equal, cv.COLOR_YUV2RGB)) ax[ii][0].axis('off') ax[ii][1].set_title('Equalized y hist') ax[ii][1].hist(y_hist_equal, 50)equalize(0, 1)equalize(0.33, 2)equalize(0.66, 3)equalize(1, 4)plt.show() 所谓的直方图均衡化，其实就是让y值多的点越来越多，y值少的点越来越少，穷者越穷，富者越富，对比度就会越来越大，区别就越大
转载地址：http://zhfqz.baihongyu.com/
你可能感兴趣的文章
mysql generic安装_MySQL 5.6 Generic Binary安装与配置_MySQL
查看>>
Mysql group by
查看>>
MySQL I 有福啦，窗口函数大大提高了取数的效率！
查看>>
mysql id自动增长 初始值 Mysql重置auto_increment初始值
查看>>
MySQL in 太多过慢的 3 种解决方案
查看>>
Mysql Innodb 锁机制
查看>>
MySQL InnoDB中意向锁的作用及原理探
查看>>
MySQL InnoDB事务隔离级别与锁机制深入解析
查看>>
Mysql InnoDB存储引擎 —— 数据页
查看>>
Mysql InnoDB存储引擎中的checkpoint技术
查看>>
Mysql InnoDB存储引擎中缓冲池Buffer Pool、Redo Log、Bin Log、Undo Log、Channge Buffer
查看>>
MySQL InnoDB引擎的锁机制详解
查看>>
Mysql INNODB引擎行锁的3种算法 Record Lock Next-Key Lock Grap Lock
查看>>
mysql InnoDB数据存储引擎 的B+树索引原理
查看>>
mysql innodb通过使用mvcc来实现可重复读
查看>>
mysql interval显示条件值_MySQL INTERVAL关键字可以使用哪些不同的单位值？
查看>>
Mysql join原理
查看>>
mysql order by多个字段排序
查看>>
MySQL Order By实现原理分析和Filesort优化
查看>>
mysql problems
查看>>