小波阈值去噪联合数学形态学的肺部图像边缘检测

ON LUNGS IMAGE EDGE DETECTION BASED ON WAVELET THRESHOLDING DENOISING AND MATHEMATICAL MORPHOLOGY

针对肺部图像的噪声问题,采用基于小波变换的阈值去噪方法去噪。在数学形态学边缘检测的基础上,选取适合肺部图像的全方位和多尺度结构元素,采用改进的形态学边缘检测算子对去噪前后的图像进行边缘检测,并给出MATLAB软件编程实现方法和核心程序。最后将所提算法对去噪前后的图像边缘检测结果进行比较。结果显示去噪后图像的峰值信噪比(PSNR)和均方误差MSE都比去噪前有明显改善,表明采用的算法不但能有效地去除噪声,同时还能保留边缘的细节,检测出更光滑、清晰的肺部图像边缘。结果也证明了小波阈值去噪联合数学形态学对肺部病灶图像进行边缘检测的有效性。

Aiming at the problem of noise in lung image, we use wavelet transform-based threshold denoising method to eliminate the noise. Then on the basis of mathematical morphology edge detection, by choosing the omnidirectional and multi-scale structural elements fit-ting the lung image, and using the improved morphological edge detection operators, we carry out edge detection on the images with noise and after denoising, and provide the implementation method and core program with MATLAB software programming. At last, we compare the pro- posed algorithm with the edge detection results of noisy image and denoised image. The results show that the peak signal-to-noise ratio （PSNR） and the mean squared error （MSE） of the denoised image have a significant improvement than the noisy image has, this illustrates that the algorithm used in this paper can effectively remove the noise while preserving the edge detail, and can detect the lung image with smoother and clearer edges. This also proves that the method of wavelet threshold denoising in conjunction with mathematical morphology is effective in edge detection of the lung lesions image.