摘要图像做为世界上一种重要的信息源，人类可以在其中提取到许多信息。图像处理逐渐发展成为了一门重要的学科，许多科学先辈提出了多种图像处理的方法，直方图处理图像是其中一种较为常用的方法。传统直方图处理方法，对于图像对比度的增强有着一定的缺点。该种增强方法对于图像不加以选择，对于有多个峰值，亮度不是极度不均衡的图像，这种方法会使得图像丢失大量的细节，引入不必要的恶化情况，同时出现对比度不正确。这种方法对于视频产品中也不适合实现。为了适用于视频产品，同时保证视频图像的对比度正确性，本文着重的论述了一种适用于视频图像对比度的直方图处理方法，即亮度保持的动态直方图均衡化图像对比度增强方法（BPDHE）。这种方法可以产生出图像平均亮度几乎等于输入图像的平均亮度的图像，以此来保持平均亮度不变以适应视频产品。首先，将图像通过高斯滤波器滤出噪声，再做出图像灰度频率直方图。对其直方图进行一维滤波，平滑直方图形成光滑曲线，选取该曲线的峰值。将该直方图基于选取的最大值进行分割，映射到一个新的动态范围。分别对这些新的动态范围进行直方图均衡化处理。因为直方图均衡化的处理，还有对于新动态范围的改变，每个直方图的平均亮度肯定会发生改变。因此，该方法的最后一步是将输出图像的亮度进行规范化处理。经过试验，该方法很好的保留原图像的平均亮度，同时做到了对比度增强，使图像部分细节更加凸显，改善了图像的视觉体验。70184

毕业论文关键词： 图像直方图均衡化 视频图像 保持亮度 图像对比度增强 BPDHE

Abstract Image as an important source of information in the world, where humans can extract a lot of information. Image processing has gradually developed into an important discipline, many scientific predecessors proposed a variety of image processing methods, histogram processing image is one of the more commonly used method. Traditional histogram processing methods, for the enhancement of image contrast has some shortcomings. This method does not allow the image to be selected. For images with multiple peaks and brightness that are not extremely unbalanced, this method will make the image lose a lot of detail, introduce unnecessary deterioration, and the contrast is not correct. This method is not suitable for video products. In order to apply to the video product, and to ensure the correctness of the contrast of the video image, this paper focuses on a histogram processing method suitable for video image contrast, that is, the dynamic histogram equalization image contrast enhancement method (BPDHE). This method can produce an image in which the average luminance of the image is almost equal to the average luminance of the input image, thereby maintaining the average luminance unchanged to accommodate the video product. First, the image through the Gaussian filter filter out the noise, and then make the image gray frequency histogram. One square filter of its histogram, smooth histogram to form a smooth curve, select the peak of the curve. Divide the histogram based on the selected maximum value and map to a new dynamic range. Respectively, these new dynamic range of histogram equalization processing. Because of the histogram equalization process, and for the new dynamic range changes, the average brightness of each histogram will certainly change. Thus, the last step of the method is to normalize the brightness of the output image. After the experiment, the method is very good to retain the average brightness of the original image, while doing the contrast enhancement, the image part of the more prominent, improve the visual experience of the image.

Keywords: image histogram equalization video image hold brightness image contrast enhancement BPDHE