基于改进细菌觅食优化的彩超图像增强算法

Color Doppler Image Enhancement Algorithm Based on Improved Bacterial Foraging Optimization

受彩色多普勒超声影像工作设备的限制,彩超图像作为医(疗)学辅助证据一般采用传统的算法(诸如空域增强中的直方图变换或频域增强中的小波变换)进行增强,而直方图变换后的彩超图像失真度过大,小波变换的参数调试更为繁琐。为此,本文提出了一种改进细菌觅食优化的图像增强算法。首先将彩超图像转到HSV颜色空间中,而后利用熵视觉显著性机制对亮度分量进行处理,以便获得显著性点集,将其作为细菌觅食优化初始菌群的候选集合,根据随机佳点理论抽选出合适的菌群点集后,对不完全Beta函数中的两个调节因子进行寻优,确定最优参数,最后将增强后的亮度分量整合色调分量和饱和度分量,并转回到RGB颜色空间中,得到增强后的彩超图像。文中的数据来源于某医院彩超诊断室,经MATLAB仿真测试后,证明所提算法同模因及杜鹃优化算法相比,峰值信噪比平均提高了7.29分贝,且具有良好的主观分辨效果。

Due to the limitation of color Doppler ultrasound imaging equipment,color Doppler ultrasound images should be enhanced if they become medical(therapeutic)auxiliary evidences.The enhancement algorithms are always histogram transformation based on spatial domain enhancement or wavelet transformation based on frequency domain enhancement.However,the distortion of color ultrasound image after histogram transform is too large,and the parameter debugging of wavelet transform is more tedious.As for that,an image enhancement algorithm of improved bacterial foraging optimization is proposed in our manuscript.Firstly,the color Doppler ultrasound image is transferred to HSV color space.Then the luminance component is processed by entropy-based visual saliency mechanism in order to obtain the salient point set,which is used as the candidate set for the initial population of bacterial foraging optimization.After the suitable population point set is selected according to the random good point theory,the two adjustable factors in the incomplete Beta function will be optimized till the optimal parameters are determined.Finally,the enhanced brightness component is integrated with hue component and saturation component to form improved HSV image.Then transfer it back to RGB color space to get the enhanced color ultrasound image.The data in this paper comes from the diagnostic room of color Doppler ultrasound in a hospital.After MATLAB simulation test,it is proved that compared with MEME and cuckoo optimization algorithm,the peak signal-to-noise ratio(PSNR)of the proposed algorithm is improved by 7.29 decibels on average,and has a good subjective resolution effect.