基于Retinex和SURF的医学图像配准与拼接

发布时间：2018-04-13 16:27

本文选题：图像拼接 + 多尺度Retinex　；参考：《计算机应用研究》2017年10期

【摘要】：针对医疗成像设备拍摄范围受限,以及医学图片受电子噪声和射线本身易散射、照度不均等复杂因素影响干扰医生诊断病情的问题,提出了基于Retinex和SURF的医学图像配准与拼接算法。算法首先通过三边滤波器改进多尺度Retinex方法增强图像,利用相位相关互功率谱快速估计重叠区域,减少配准范围;接着通过SURF局部特征提取算法提取特征点并精确提纯,建立图像变换矩阵实现拼接;最后针对接缝明显的现象,利用提升小波变换多分辨率融合算法进一步对图像进行融合处理,使得图像接缝过渡平滑自然。实验及质量评价标准的结果证实,新方法处理的医学图像配准与拼接对复杂环境有较高的鲁棒性。
[Abstract]:According to the medical imaging equipment shooting range is limited, and medical images by electronic noise and radiation itself to scattering, Illumination Uneven complicated factors interfere with doctors to diagnose the disease problems, proposes the medical image registration and stitching algorithm based on Retinex and SURF. Firstly, through the three edge filter improved multi-scale Retinex image enhancement method, using phase cross power spectrum estimation to reduce the overlap area, registration scope; and then using the SURF feature extraction algorithm to extract the feature points accurately and purification, the image transform matrix to achieve the most in seam stitching; obvious phenomenon, using lifting wavelet transform multi-resolution fusion algorithm of image fusion processing, the image joint is smooth and natural. The experimental results and the quality evaluation criteria that the new processing method of medical image registration and stitching in complex environment It has high robustness.

【作者单位】：徐州工程学院信电工程学院;徐州市机器视觉应用工程技术研究中心;
【基金】：江苏省产学研联合创新资金资助项目(BY2013020) 江苏省高校自然科学研究面上项目(15KJB520033,16KJB510022) 江苏省自然科学基金青年基金资助项目(BK20160966)
【分类号】：TP391.41

【相似文献】