多尺度下低对比度红外热图像增强技术

THE ENHANCEMENT TECHNOLOGY OF LOW CONTRAST INFRARED IMAGE UNDER MULTISCALE

对红外热图像进行增强时，常采用基于直方图的图像灰度线性或非线性变换技术以及中值滤波方法，使图像的目标与背景之间的灰度差别放大并且滤除噪声，从而达到增强图像的目的，其缺点是灰度变换技术易放大噪声而中值滤波方法只能滤除高斯白噪声。由于梯度反映了图像灰度之间的差异，而根据二进小波变换可以建立红外热图像的梯度矢量图，通过增强红外热图像梯度的幅度，可以有效地增强图像对比度。根据脉冲噪声和高斯白噪声的多尺度下小波变换特性，抑制噪声点的梯度且增加目标的梯度幅度，能达到既增强图像又抑制多种噪声的目的。

In order to enhance low contrast infrared image, histogram - based linear or nonlinear transform technology on tonal value of image and median filter are often used to amplify the difference between object and background and to reduce the noise.But the histogram - based technofogy may amplify the noise in infrared image and the median filter can only reduce Gauss white noise. In this paper, a method of dyadic wavelet transform to get the gradients of the infrared image is given. The contrast of the low contrast infrared image can be effectively enhanced by transforming the gradients of the image. The characteristics of the impulse noise and Gauss white noise can also obtained by multiscale wavelet transform, so that the noise of the infrared image is controled and the image is enhanced.