摘要
原木在堆放时其端面不在一个平面,自然光线下,凹进去的原木很容易陷入阴影中。为了准确分割原木端面图像,解决阴影消除的问题,提出了模式识别的方法。首先,提出了自适应阈值的合并解决均值漂移分割参数选择不合适引起的过分割和过合并问题;再将30维的颜色直方图和局部二进制模式(LBP)纹理直方图作为特征量,利用随机森林分类器进行训练和预测,将区域图像分为端面、阴影、背景3类;最后,利用图像增强算法进行阴影消除。对阴影消除前后的端面图像作分割,对比结果显示了此算法的有效性。
In the natural environment,the end surface of the log is not in a plane,and concave logs always fall into the shadows.In order to accurately segment the log ends surface and solve the problem of shadow elimination,the method of pattern recognition was proposed.Adaptive threshold merging method was proposed to solve the over segmentation and over merging problem caused by inappropriate meanshift segmentation.Then,30 dimensional color histogram and LBP texture histogram were used as feathers,and a random forest classifier was used for training and prediction,the regional images was divided into face,shadow and background.Finally,the image enhancement algorithm was used for shadow elimination.
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2016年第8期92-96,共5页
Journal of Northeast Forestry University
基金
黑龙江省青年科学基金项目(QC2015080)
关键词
自适应分割
阴影检测
阴影消除
Adaptive segmentation
Shadow detection
Shadow elimination
