基于改进的快速最大熵多阈值的成熟草莓图像分割  

Ripe Strawberry Image Segmentation Based on Improved Fast Maximum Entropy Multiple Threshold

作　　者：阮松[1] 孙开琼[2] 覃磊[1,3] 张文质[1] 周康[1] 

机构地区：[1]武汉轻工大学数学与计算机学院,武汉430023 [2]南昌航空大学测试与光电工程学院,南昌330063 [3]华中科技大学自动化学院,武汉430074

出　　处：《东北农业科学》2016年第1期107-112,共6页Journal of Northeast Agricultural Sciences

基　　金：国家自然科学基金资助项目(61163046);湖北省教育厅科技计划项目(B20091803);武汉轻工大学科研计划项目(2015y03);武汉轻工大学校立大学生科研项目(xsky2015032)

摘　　要：成熟草莓图像分割是草莓机械化采摘中首要解决的难题之一,最大熵多阈值算法是图像分割中效果较稳定的方法之一,但存在计算复杂度高、分割速度慢等缺点。为降低算法的计算复杂度、加快搜索速度,提出了一种改进的快速最大熵多阈值图像分割算法(IFMEMT)。首先提取RGB彩色图像的R分量灰度图像及灰度图像信息,然后应用IFMEMT算法求得最大熵及对应的阈值,最后进行图像分割。实验结果表明,在复杂环境下IFMEMT相对OTSU等图像分割算法不仅能达到相同甚至更好的分割效果,而且有更好的分割效率,能满足成熟草莓机械化采摘的实时性要求。Ripe strawberry image segmentation is one of the primary problems of strawberry mechanization picking. Maximum entropy multiple threshold algorithm is one of the stability threshold methods in the image segmentation field, but it has the shortcomings of high computational complexity and slow segmentation speed and so on. To reduce its computational complexity and accelerate its search speed, an improved fast maximum entropy multiple threshold strawberry image segmentation method （IFMEMT） was proposed in the paper. Firstly, the R component gray images of RGB color images and their gray image information were extracted, and then applied IFMEMT algorithm, the maximum entropy and their corresponding thresholds were obtained; finally, the images were segmented. The experimental results showed that IFMEMT in a variety of complex environments could not only achieve the same or even better segmentation effect than OTSU algorithm, but also had better segmentation efficiency, and it could meet the real-time requirement of ripe strawberry mechanization picking.

关 键 词：图像分割 快速 最大熵 多阈值 

分 类 号：S126[农业科学—农业基础科学]