【摘要】 在仔细研究白细胞图像特点的基础上,提出了二叉树分类算法,主要提取细胞的色彩参数以及核的形状参数,用于判决识别。在细胞抽提阶段,只进行核的抽取,避免了细胞浆与背景颜色接近,难以分离而导致的因细胞浆提取不完整引起的判决误差,提高了系统的识别率;克服了统计模式识别算法中需要大量训练样本和提取众多特征参数的缺点,大大减少了矩阵运算时间。更多还原

【Abstract】 Based on the study of leukocyte images, the algorithm of two-prong tree classification was proposed. The cell color-parameters and nucleus shape-parameters were extracted for the decision and recognition.Only nucleuses were extracted so that the decision errors were avoided,resulting from the incomplete extraction of cytoplasm due to the color similarity between cytoplasm and background.The shortcomings of requiring many training samples and feature parameters in statistical pattern recognition algorithm were overcome. The matrix operation time was reduced.更多还原