Lab分通道直方图的彩色图像分割算法及应用

doi:10.7517/issn.1674-0475.181004

摘要/Abstract

摘要： 本文针对传统灰度直方图分割法未综合考虑图像色度及纹理特征、对灰度差异不明显或灰度范围重叠的图像出现过分割或欠分割等问题，提出一种新的基于Lab分通道直方图的彩色图像分割算法，引入具有序列不相关性的亮度L通道、红绿a通道及蓝黄b通道3种分割依据，通过Newton插值法进行拟合运算，可针对不同亮度、色度属性图像进行自由选择，并运用邻域灰度值相匹配原则解决相邻目标区域边缘像素的准确匹配问题，分局部、分形态、分区域实现图像中不同目标的提取。经验证，该法对区域亮度差异较大图像及区域色度差异显著于亮度差异图像的分割效果，均优于传统灰度直方图分割法，极大提升了直方图分割算法的适用性。将其与经典Reinhard色彩迁移算法结合，将源图像感兴趣目标区域经分通道分割后分别进行色彩迁移变换，较好解决了经典Reinhard算法对图像非目标区域的干扰、色彩误传及阶调层次损失严重等问题，突破传统迁移算法只能整体着色的局限性，实现分区域精准着色。

Abstract: In view of the color and texture features of the image are not taken into account in the traditional gray scale histogram segmentation algorithm, this paper proposed a new color image segmentation algorithm based on Lab sub channel histogram to solve the problems such as over segmentation or under segmentation for no obvious gray value difference images or grayscale overlapped images. The algorithm introduces three segmentation bases with sequence irrelevance:luminance "L" channel, red-green "a" channel and blue-yellow "b" channel, and then use the Newton interpolation method to fit operation. Users can choose the channel freely for different brightness and chromaticity attribute images. The paper also uses the matching principle of adjacent area gray value to solve the exact matching problem of pixels in the adjacent target area. The algorithm realizes the extraction of different objects in the image by sub office, sub morphology and subarea. In the light of images with significant difference in regional luminance characteristics or regional color difference greater than regional brightness difference, the segmentation results of the algorithm in this paper are better than the traditional gray scale histogram segmentation algorithm through a series of experimental verification, it greatly improves the applicability of histogram segmentation algorithm. Then we combine the new algorithm with the classic Reinhard color transfer algorithm, use the sub channel segmentation algorithm to obtain the target regions of interest from the source image and then transfer them separately. It solves problems of the interference to non-target area of image, color miscommunication and serious level loss of classical Reinhard color transfer algorithm. It breaks through the limitation that the traditional transfer algorithm can only coloring the whole region, and realizes accurate coloring in the subarea.

Key words: image processing, sequence irrelevance, Lab sub channel histogram segmentation algorithm, Newton interpolation method, matching principle of adjacent area gray value, evenness evaluation index, Reinhard color transfer algorithm

于艺铭, 金典, 王琪, 张琪, 陈茜, 王小菊. Lab分通道直方图的彩色图像分割算法及应用[J]. 影像科学与光化学, 2019, 37(1): 18-32.

YU Yiming, JIN Dian, WANG Qi, ZHANG Qi, CHEN Xi, WANG Xiaoju. Color Image Segmentation Algorithm Based on Lab Sub Channel Histogram and Its Application[J]. Imaging Science and Photochemistry, 2019, 37(1): 18-32.

参考文献

[1] 宋熙煜,周利莉,李中国,陈健,曾磊,闫镔.图像分割中的超像素方法研究综述[J].中国图象图形学报,2015,20(5):599-608. Song X Y, Zhou L L, Li Z G, Chen J, Zeng L, Yan B. Review of super-pixel methods in image segmentation[J]. Journal of Image Graphics, 2015, 20(5):599-608.
[2] 文政颖,于海鹏.基于多Gamma分布模型的SAR图像直方图分割算法[J].计算机工程与设计,2014,35(6):2104-2108. Wen Z Y, Yu H P. SAR image histogram segmentation algorithm based on multi-Gamma distribution model[J].Computer Engineering and Design, 2014, 35(6):2104-2108.
[3] 郑毅,郑苹.基于模糊熵和模拟退火算法的双阈值图像分割[J].电子测量与仪器学报,2014,28(4):360-367. Zheng Y, Zheng P. Double threshold image segmentation based on fuzzy entropy and simulated annealing algorithm[J].Journal of Electronic Measurement and Instrument, 2014, 28(4):360-367.
[4] 刘霁雨,龚敬,聂生东.利用测地线距离直方图分割血管粘连型肺结节的算法[J].计算机辅助设计与图形学学报,2017,29(3):493-500. Liu J Y, Gong J, Nie S D. An algorithm for segmentation of adhesively vascular pulmonary nodules by geodesic distance histogram[J].Journal of Computer Aided Design and Graphics, 2017, 29(3):493-500.
[5] 越琳,高丽燕.直方图分割耦合裁剪控制均衡化的图像增强算法[J].计算机工程与设计,2017, 38(2):465-469. Yue L, Gao L Y. Image enhancement algorithm for histogram segmentation coupled with cropping control equalization[J].Computer Engineering and Design, 2017, 38(2):465-469.
[6] 吴晓东,冯伍法,张继领,王康康,务宇宽.基于多峰灰度直方图分割的遥感影像码头目标提取[J].影像技术,2016, 28(2):46-47. Wu X D, Feng W F, Zhang J L, Wang K K, Wu Y K. Target extraction of remote sensing image wharf based on multi-peak gray histogram segmentation[J].Image Technology, 2016, 28(2):46-47.
[7] 王卫星,田利平,王悦.基于改进的图论最小生成树及骨架距离直方图分割细胞图像[J].光学精密工程,2013,21(9):2464-2472. Wang W X, Tian L P, Wang Y. Cell image segmentation based on improved graph theory minimum spanning tree and skeleton distance histogram[J].Optical Precision Engineering, 2013, 21(9):2464-2472.
[8] 赵向文,靳国旺,许竞轩,刘继东.基于多阈值大津法直方图分割的SAR数据伪彩色量化方法[J].测绘科学技术学报,2012,29(6):445-449. Zhao X W, Jin G W, Xu J X, Liu J D. A pseudo-color quantization method for SAR data based on multi-threshold Otsu histogram segmentation[J].Journal of Surveying and Mapping Science and Technology, 2012, 29(6):445-449.
[9] 李炜. 群体目标动态分析研究[D].济南:山东大学,2011. Li W. Dynamic analysis of group targets[D]. Jinan:Shandong University, 2011.
[10] 胡琼,汪荣贵,胡韦伟,杨万挺.基于直方图分割的彩色图像增强算法[J].中国图象图形学报,2009,14(9):1776-1781. Hu Q, Wang R G, Hu W W, Yang W T. Color image enhancement algorithm based on histogram segmentation[J].Chinese Journal of Image Graphics, 2009, 14(9):1776-1781.
[11] 刘金,金炜东.三维自适应最小误差阈值分割法[J].中国图象图形学报,2013, 18(11):1416-1424. Liu J,Jin W D.Three-dimensional adaptive minimum error threshold segmentation[J].Chinese Journal of Image Graphics, 2013, 18(11):1416-1424.
[12] 安玮,徐晖,李宏,孙仲康.基于奇异性分析的图像直方图分割算法[J].模式识别与人工智能,1999, 12(1):13-18. An W, Xu H, Li H, Sun Z K. Image histogram segmentation algorithm based on singularity analysis[J].Pattern Recognition and Artificial Intelligence, 1999, 12(1):13-18.
[13] 吴掬鸥,袁晓桂.基于阈值分割技术的图像分割法研究[J].现代电子技术,2016, 39(16):105-107. Wu J O, Yuan X G. Image segmentation based on threshold segmentation[J].Modern Electronic Technology, 2016, 39(16):105-107.
[14] 项衍,叶擎昊,刘建国,张天舒,范广强.基于图像边缘检测法反演大气边界层高度[J].中国激光,2016, 43(7):191-197. Xiang Y, Ye Q H, Liu J G, Zhang T S, Fan G Q.Retrieval of atmospheric boundary layer height based on image edge detection[J].China Laser, 2016, 43(7):191-197.
[15] 徐少平,刘小平,李春泉,胡凌燕,杨晓辉.基于区域最大相似度的快速图像分割算法[J].光电子·激光,2013, 24(5):990-998. Xu S P, Liu X P, Li C Q, Hu L Y, Yang X H. Fast image segmentation algorithm based on region maximum similarity[J].Optoelectronics·Laser, 2013, 24(5):990-998.
[16] 裴继红,谢纹信.直方图模糊约束FCM聚类自适应多阈值图像分割[J].电子学报,1999,27(10):38-42. Pei J H, Xie W X. Histogram fuzzy constrained FCM clustering adaptive multi-threshold image segmentation[J].Journal of Electronics, 1999,27(10):38-42.
[17] 冯惠扬,王中康,王宏伟.人脸定位算法的研究[J].计算机工程与应用,2003,39(30):101-120. Feng H Y, Wang Z K, Wang H W. Research on face location algorithm[J].Computer Engineering and Application, 2003,39(30):101-120.
[18] Liao P S, Cheu T S, Chung P C. A fast algorithm for multilevel thresholding[J].Journal of Information Science and Engineering, 2001,17(5):713-727.
[19] 王文宁,王汇源,牟文英.一种新的灰度直方图分割阈值的自动检测方法[J].计算机工程与应用,2005, (26):89-90-173. Wang W N, Wang H Y, Mu W Y.A new automatic threshold detection method for gray histogram segmentation[J].Computer Engineering and Application, 2005, (26):89-90-173.
[20] 龙建武,申铉京,陈海鹏.自适应最小误差阈值分割算法[J].自动化学报,2012, 38(7):1134-1144. Long J W, Shen X J, Chen H P. Adaptive minimum error threshold segmentation algorithm[J].Journal of Automation, 2012, 38(7):1134-1144.
[21] 张弘,范九伦.二维Arimoto熵直线型阈值分割法[J].光子学报,2013, 42(2):234-240. Zhang H, Fan J L. Two dimensional Arimoto entropy linear threshold segmentation method[J].Acta Photonics, 2013, 42(2):234-240.
[22] 王兴玲.最大类间方差车牌字符分割的模板匹配算法[J].计算机工程, 2006, (19):193-195. Wang X L.Template matching algorithm for character segmentation of license plate with maximum inter-class variance[J].Computer Engineering, 2006, (19):193-195.
[23] Takahiko H. Colorization algorithm using probabilistic relaxation[J]. Image and Vision Computing,2004, 22(3):197-202.
[24] Reinhard E,Ashikhmin M,Gooch B,Shirley P.Color transfer between images[J].IEEE Computer Graphics and Applications,2001, 21(5):34-41.
[25] Tomohisa T,Takahiko H,Hiroaki K.Improvement on colorization accuracy by partitioning algorithm in CIELAB color space[J]. PCM 2004,794-801.