基于道路监控的图像增强算法研究

摘要

To increase the human face recognition rate of road monitoring system under the condition of changing illumination,a new approach for image enhancement based on a fusion of multi-scale Retinex and contrast limited adaptive histogram equalization(CLAHE) is introduced in this paper.First,we remove the noise of image and make a process of CLAHE,which can increase the contrast of local pixels.Then the multiscale Retinex is implemented to optimize the details in the dark areas.Finally,a weigeted fusion of multiscale Retinx and CLAHE is made according to their own characters.The experimental results on a MATLAB R2015b simulation platform evaluate the performance of the performance of the proposed fusion algorithm in terms standard deviation,average gradient and so on,indicating that this new algorithm is practicable and advantaged in image enhancement.%为了提高道路监控系统在光照条件变化状况下的人脸识别率,针对采集到低照度彩色图像提出一种将多尺度Retinex算法和限幅自适应直方图均衡化(CLAHE)融合的图像增强算法.首先对含有噪声图像去噪后进行CLAHE处理,可增强图像局部像素的对比度,再通过多尺度Retinex图像增强可有效提高暗区域的细节,最后根据两种算法特点进行加权融合.经过matlab2015b平台仿真测试得到很好的视觉效果,图像的标准差和平均梯度等指标得到了提升,验证了融合算法对图像增强的独特优越性以及算法可行性.