计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (21): 195-202.DOI: 10.3778/j.issn.1002-8331.1604-0369

• 图形图像处理 • 上一篇    下一篇

基于特征分类的麻将图像光照不均校正方法

王财盛1,朱  威1,2,胡文文1,徐鑫涛1   

  1. 1.浙江工业大学 信息工程学院,杭州 310023
    2.浙江省嵌入式系统联合重点实验室,杭州 310023
  • 出版日期:2017-11-01 发布日期:2017-11-15

Uneven illumination correction of mahjong image based on feature classification

WANG Caisheng1, ZHU Wei1,2, HU Wenwen1, XU Xintao1   

  1. 1.College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, China
    2.United Key Laboratory of Embedded System of Zhejiang Province, Hangzhou 310023, China
  • Online:2017-11-01 Published:2017-11-15

摘要: 针对麻将排版视觉检测装置采集到的图像存在光照不均问题,提出了一种基于麻将特征分类的自适应图像校正方法。通过对麻将的排版特征进行分析,采用背景差分法和投影法将整体排版麻将灰度图从采集图像中分割出来;将整体排版麻将划分为多个单个麻将,并通过对单个麻将的光照特征进行分析,将单个麻将按光照类型分为均匀类麻将、阴影类麻将和高光类麻将共三类;对阴影类麻将和高光类麻将进行均值归一化处理,再对麻将图像进行整体锐化,并结合色度分量增强麻将的图案特征;根据麻将噪声特征,基于空域幅值统计去除阴影类麻将中的图像噪点。实验结果表明,与现有方法相比,该方法能更好地改善麻将图像光照不均现象,处理后的图像有效地提高了麻将排版识别率。

关键词: 麻将图像, 光照不均, 图像锐化, 色度分量, 局部去噪

Abstract: To solve the uneven illumination of mahjong image for the visual detection of mahjong arrayal, an adaptive correction method of mahjong image based on the feature classification is proposed. Firstly, according to the feature analysis of mahjong’s arrangement, the background subtraction and projection methods are used to segment the whole mahjong image from the capture image. Secondly, the whole mahjong is divided into single mahjong, which is further divided into three types:highlight mahjong, shadow mahjong and uniform mahjong by the feature analysis of illumination. Thirdly, mean normalization is performed for the highlight mahjong and shadow mahjong. Fourthly, the whole mahjong image is sharpened, and its characteristics are strengthened by using the chrominance component. Finally, it uses the local method of analyzing the amplitude around to remove the noise of the shadow mahjong based on the feature of the noise. Experimental results show that the proposed method has better correction for the uneven illumination of mahjong image than state-of-art methods and can improve the recognition rate effectively for the mahjong arrayal detection.

Key words: mahjong image, uneven illumination, sharpening, chrominance, local denoising