基于三维虚拟技术的夜间低照度激光图像增强

Night low illumination laser image enhancement based on 3D virtual technology

以提升夜间低照度激光图像的清晰度与分辨率为目的,提出基于三维虚拟技术的夜间低照度激光图像增强方法。利用多尺度主特征提取法确定夜间低照度激光图像的照射分量,并对照射分量进行优化;利用优化后的照射分量计算夜间低照度激光图像的反射光;针对反射光图像进行强光及噪声抑制,设置响应权值将原夜间低照度激光图像与抑制后的反射光图像进行加权合成,得到视觉特征增强图像。对视觉特征增强图像进行细节特征增强,利用三维虚拟技术对图像进行变换,可变换激光图像局部,不仅提升激光图像局部复杂性,同时增强激光图像细节特征描述;同时利用分割重组技术提升图像真实性。测试结果显示采用该方法增强后,夜间低照度激光图像的峰值信噪比和机构相似度指数均较高,均值分别为35.04 dB和0.90。

In order to improve the clarity and resolution of night low illumination laser image,a method of night low illumination laser image enhancement based on 3 D virtual technology is proposed. The multi-scale principal feature extraction method is used to determine and optimize the illumination component of the night low-intensity laser image.The optimized illumination component is used to calculate the reflected light of the low-intensity laser image at night;the strong light and noise are suppressed for the reflected light image,and the response weight is set to combine the original low-intensity laser image and the suppressed reflected light image. The visual feature enhancement image is obtained. In this paper,the detail feature enhancement of the visual feature enhancement image is carried out. The 3 D virtual technology is used to transform the image to transform the local part of the laser image,which not only improves the local complexity of the laser image,but also enhances the detail feature description of the laser image;at the same time,the segmentation and reorganization technology is used to enhance the authenticity of the image. The test results show that the peak signal-to-noise ratio( PSNR) and mechanism similarity index( CSI) of night low illumination laser images are both high,with the average values of 35. 04 db and 0. 90,respectively.