中文摘要：

血管内超声图像同时具有低对比度和低信噪比的特征，传统的图像增强算法会明显放大噪声．为此，提出一种基于边缘表示的算法来增强血管内超声图像的对比度，采用平滑函数的导数作为小波函数对图像进行二进小波分解，使用小波系数在各尺度间的局部极大值来刻画图像的多尺度边缘，通过拉伸小波系数极大值再重构图像来增强灰度图像的对比度．实验结果表明，本文提出的增强算法在有效增强图像对比度的同时避免了放大噪声，并且可以通过调节不同尺度的拉伸因子来获得不同的增强效果，同时提高了运算速度．

英文摘要：

Intravascular ultrasound image has both low contrast and signal-noise ratio, traditional en- hancement methods will amplify noise when enhancing image contrast. A contrast enhancement algo- rithm based on multiscale edges representation of image is proposed. After dyadic wavelet decomposition of images when using derivative of smooth functions as wavelet function, wavelet coefficients extrema can depict multiscale boundary of images, so image contrast can be enhanced by stretching wavelet coef- ficients extrema and reconstructing images. Experimental results show that this new algorithm is free from amplifying noise when enhancing image contrast, obtains different enhancement results by adjus- ting amplifying weights and has low run time.