高光条件下视频人像处理方法研究

摘要

ABSTRACT:ObjectiveUnder high light conditions, human video images are often blurred, resulting in the details to be lost, therefore, a study on processing methods of human video images under high light is conducted in this paper.MethodsFirstly, the problems of human video images under high light are analyzed in detail. Secondly, the lfow chart for processing human video image is presented. The methods on how to process the high light human video images are described under normal and/or infrared monitor mode. Finally, the speciifc processing methods are introduced. The human video images of high quality, less affected by high light and their key frame of larger size, are picked out. In spatial domain, curve adjustment and gamma correction have been used for enhancing the high light human images. And so have that in frequency domain with utilization of high pass and/or homomorphic ifltering.ResultsFor the high light human images video, it is necessary to enhance the original high light human images before their reconstructing and synthesizing. For facial details missing, it is better for human images to inhibit high light on comparison of the reconstructed and synthesized human images with the original ones. From the processing effect, the methods established for enhancing human images in frequency domain are suggested. Compared to the color video ones, the infrared human video images lose their color. It is indispensable for the processing to conduct curve adjustment or gamma correction before sharpening in enthancement by high pass ifltering. In some cases, the processed results by gray imaging are clearer than by color handling. Thus, conversion of the color image to grayscale one is optional sometimes.Conclusion The right way for processing high light human video images can provide more accurate clues for video investigation. The methods proposed here could improve the visual effect of high light human images.%目的：针对高光条件下导致的视频人像模糊、细节缺失的情况，研究高光条件下的视频人像的处理方法，改善视频中高光人像的视觉效果。方法详细分析了高光条件下视频人像存在的问题，给出了高光视频人像处理流程图，阐述了普通监控拍摄模式下和红外监控模式下高光人像处理方法。具体处理方法：在人像视频中挑取成像质量较好、光照影响较小、尺寸较大的关键帧图像，将人像剪裁出合适的尺寸。在空域内用曲线调节、伽马校正，在频域内用高通滤波、同态滤波等方法分别对彩色高光人像和灰度高光人像进行增强处理，然后进行人像重建、合成。结果面对高光人像细节丢失这种普遍情况，在对人像重建、合成前，对原始高光人像进行增强处理是有必要的。相对于直接对原始高光人像，先对人像进行高光抑制，再进行重建、合成处理效果更好。在处理方法上，频域内的高通滤波锐化加强效果更好。相对于彩色视频人像，红外视频人像丢失了色彩，在高通滤波前进行曲线调节或伽马校正是有必要的。在某些情况下，灰度图像的处理效果比彩色图像处理效果更好，将彩色图像转换为灰度图像然后再进行重建处理也是可行的方法。结论正确处理高光人像能够为视频侦查提供更准确线索，文中所提方法能够改善高光人像视觉效果。