摘要： 根据大鼠肝脏免疫组化彩色图像的特征，提出一种阳性产物分割算法。在任意3幅图像中选择一批样本点，通过对这些样本点的分析找到阳性像素在RGB和HSI空间中的特征，根据这些特征提取阳性产物所在区域，采用中值滤波方法去除噪声。实验结果表明，该算法约有82.9%的分割结果与病理医生目视鉴别结果一致，比现有算法更适合大鼠肝脏免疫组化彩色图像的分割。

关键词: 图像分割, 混合颜色空间, 免疫组化彩色图像, 中值滤波

Abstract: A novel method for positive area extraction is proposed according to the characteristics of immunohisechemical color images for mouse liver. Images are transformed from RGB to HSI. It is found that the positive pixels through analyzing the sample pixels in three arbitrary images. Median filtering algorithm is used to remove the wrong areas. Experimental results show that 82.9% results are in accordance with the estimation of physicians. In comparison with other methods, this algorithm is more appropriate and effective.

Key words: image segmentation, hybrid color space, immunohistochemical color image, median filtering

中图分类号: 

• TP391