基于理想图像重建的菌落检测算法
孙琪，杨兆选，鹿凯宁
基金项目：教育部博士点基金(20090032110028)；国家自然科学基金(61100124,21106095,60976001)；天津市应用基础与前沿技术研究计划项目(10JCYBJC25500)；2011 年和2010 年天津大学自主创新基金

作者简介：孙琪，（1988-），女，研究生，主要研究方向：模式识别.
（天津大学电子信息工程学院，天津 300072）
摘要：本文提出了一种基于理想图像重建的菌落检测算法。不同于传统的基于目标检测思路的菌落检测方法，该方法将菌落检测问题转化为图像分割问题，并且为了解决传统图像分割算法在自适应阈值选择上的困难，该方法通过非均匀光照偏差的消除来重建理想的菌落图像，使得重建后的理想图像中菌落对应的高亮度区域和背景对应的低亮度区域具有较高的类间差和较低的类内差，从而可以采用简单的自适应阈值选择算法进行重建理想图像的二值化和菌落的检测。大规模的实验证明了该算法的准确性和高效性。
关键词：理想图像重建；图像分割；形态学；菌落检测
Ideal Image Reconstruction-Based Bacteria Colony Detection Method
SUN Qi, YANG Zhaoxuan, LU Kaining
(School of Electronic Information Engineering,Tianjin University, TianJin 300072)
Abstract: This paper proposes a novel bacterium colony detection method based on adaptive ideal image reconstruction. Distinct from traditional object detection-based methods for bacteria colony identification, the proposed method accomplishes this problem in the way of image segmentation. To overcome the challenges in threshold selection, this method reconstructs ideal bacteria colony image by eliminating the bias of illumination. Specifically, in the reconstructed ideal image bacteria regions correspond to high intensity parts and the backgrounds lie in relatively low intensity parts. Consequently, the reconstructed image can be conveniently binaried with adaptive image segmentation methods, like Ostu’s method and so on, and bacteria colony can be easily detected. Large-scale experiments show the accuracy and efficiency of the proposed method.
Keywords: Ideal image restoration; image segmentation; morphology; bacteria colony detection