OP"问题中的(通过直方图均衡增强图像)中需要 code 帮助. 您能告诉我什么是该算法的现代替代方法

对比度拉伸是一个广义的词，就好像您添加了高级词一样，它也是与您所关注的领域相关的特定词.这里有几个链接

http://www.ece.udel.edu/~barner/courses/eleg675/papers/Enhancement%20in%20the%20Spatial%20Domain/An%20Advanced%20Contrast%20Enhancement.pdf


http://www.eyemaginary.com/Rendering/Au​​tomaticHistogramWarping.pdf

http://portal.acm.org/citation.cfm?id=1851096


http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6WDJ-499706H-3&_user=10&_coverDate=03%2F31%2F2004&_rdoc=1&_fmt=high&_orig=gateway&_origin=网关& _sort = d& _docanchor =& view = c& _searchStrId = 1688010778& _rerunOrigin = google& _acct = C000050221& _version = 1& _urlVersion = 0& _userid = 10& md5 = 5d29ebedfebe0b1e4254898

您可能需要开箱即用地考虑对比度增强不只是在处理直方图.去相关拉伸怎么样?让我尝试简单介绍一下.如果绘制光谱，即在红色，绿色和蓝色光带(RGB)中的反射率，它将形成不规则形状，但类似于椭圆形，除非图像已经形成了良好的对比度.如果它是理想的椭球，那么理论上所有波段都具有相同的信息模式并高度相关.那么，如果您更改视图，该怎么办.将轴旋转到椭圆体的主轴，然后缩放该轴以达到最大对比度.然后第二根轴垂直于它.因此，您将获得不在第一个轴上的信息.因此，如果将RGB颜色分配给这些转换后的值，则每种颜色将代表不同的信息，并且可以获得良好的对比度.从数学上讲，这种转换称为主成分分析.

Mattikalli在2001年提出了另一种称为***旋转变换"的旋转技术.他是NASA雇员.这是专业领域中使用的一种迭代技术.但是，您可以使用此数学变换，并在每个迭代步骤中计算图像的熵.收敛到无熵变化的旋转将是获得的最大信息.但是这里需要注意的是，您可能会增强噪音.因此，在使用这些技术处理嘈杂的图像之前，预防性去除噪声是理想的选择.

甚至您也可以使用色彩空间.您知道十年前我们的卫星没有传输大量数据的能力.它在低分辨率(更多波段，更多音量)下需要多光谱，而在高分辨率(即单波段)下则需要灰度.那么如何获得色彩鲜艳但高分辨率的数据呢?我们将RGB中的多光谱传输到IHS(强度，色相和饱和度).然后，我们去除灰度级的强度，并将高分辨率灰度级图像替换为强度，然后转换回RGB.结果就是我们想要的.

因此，对比度增强是一个巨大的话题.您可能会经历最新的发展，并尝试不同的方法.那将为您带来一个很好的研究论文.希望我能提供有用的简要介绍.

OP''s question. can you tell me what is the modern and alternative methods to this algorithm

Contrast stretch is a generalized word where as if you add advanced then it is a specific word related to your filed of interest. Here a few links

http://www.ece.udel.edu/~barner/courses/eleg675/papers/Enhancement%20in%20the%20Spatial%20Domain/An%20Advanced%20Contrast%20Enhancement.pdf


http://www.eyemaginary.com/Rendering/AutomaticHistogramWarping.pdf

http://portal.acm.org/citation.cfm?id=1851096


http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6WDJ-499706H-3&_user=10&_coverDate=03%2F31%2F2004&_rdoc=1&_fmt=high&_orig=gateway&_origin=gateway&_sort=d&_docanchor=&view=c&_searchStrId=1688010778&_rerunOrigin=google&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=5d29ebedfebe0b1e4254898d681493b4&searchtype=a


You may need to think out of box that contrast enhancement is not just playing with histograms. How about a decorrelation stretch?. Let me try to give a brief introduction. If you plot the spectral i.e the reflectance in the red, green and blue bands of light (RGB) it forms a irregular shaped, but resemble to an ellipsoid, unless the image is already in good contrast. If it is an ideal ellipsoid then theoretically all the bands has same patterns of information and highly correlated. So what if you change your viewings. Rotate the axis to the major axis of the ellipsoid, then the axis is scaled to maximum contrast. Then second axis perpendicular to it. So you will be getting information which are not in the first axis. So if you assign RGB colors to these transformed values each color will represent different information and you get a good contrast. Mathematically this transformation is called principal component analysis.

There is another rotational technique called "Optimal rotational transformation" presented by Mattikalli in 2001. He is NASA employee. This is an iterative technique used in a specialized field. However you can use this mathematical transformation and in each iterative step calculate your image''s entropy. The rotation which converges to no entropy changes would be the maximum information obtained. But need to be care here may be you will be enhancing the noises. So precautionary noise removal is ideal before doing these techniques for noisy images.

Even you can play with color spaces. You know 10 years back our satellite don''t have the capacity of transfer large volume of data. It takes multispectral in a low resolutions (more bands, more volume), but the gray scale in high resolution i.e single band. So how about getting a colorful but high resolution data? We transfer the the multispectral in RGB to an IHS (intensity, Hue and saturation). Then we remove the intensity which is on grey scale and put our high resolution gray scale image in the place of intensity and transform back to RGB. It results what we want.

So contrast enhancement is vast topic. You may go through recent developments and experiment with different methods. That will result in a good research thesis for you. I hope I given a useful brief introduction.