摘要
提出了一种基于多级灰度差特征的脂肪肝B超图像识别方法。在B超图像中提取一个梯形区域,将该区域分成n×m个小梯形;并计算每个小梯形区域的平均灰度值,得到从近场到远场的灰度差矩阵作为多级灰度差特征;结合人工神经网络进行脂肪肝识别。实验结果表明,新方法取得了较好地识别效果,提高了B超图像脂肪肝的识别率。
A novel method for fatty liver B-scan image recognition with multi-grade gray level difference is proposed.First a trapeziform area is extracted from B-scan image.And then the trapeziform area is divided into n×m regular trapeziform areas.After computed every small areas average gray level,the multi-grade gray level difference axis matrix is gained from near area to far area.Finally the image is recognized with neural network based on multi-grade gray level difference.The experiment shows that new method gets a good effect in fatty liver B-scan images recognition.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第5期1899-1903,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(60272099)
柳州市科学研究与技术开发计划基金项目(柳科综字[2007]11号)
关键词
脂肪肝
B超图像
特征提取
多级灰度差
人工神经网络
fatty liver
b-scan image
feature extraction
multi-grade gray level difference
artificial neural network
