二值化是逐像素处理,而逐像素处理会有很多效果,这主要是给人眼看的,因为像素值的变化,直观的就是图像变化,比如颜色。
颜色变幻处理
OpenCV提供了一些图片,如下:
粗看是一些风格,细看一下,其实是各行颜色一致,看属性知道图像尺寸为256 * 30。
256是字节宽度,也是各通道取值范围。
所以,可以用查表法直接替代处理,即将上面一张图片作为输入,则各行就将原值的0-255映射成为图片相应列的值。图像尺寸设置为30,其实也无所谓多少行,至少一行就行,不过行数太少了,人看起来就感觉,行数多了,读到计算机内存里面处理也浪费。那就折衷选个30吧。
代码处理,这就看水平了,可以一次性全读入,也可以只读取需要的图片。
QString colorMapPath = THelper::File::GetPath_Resource() + "colorMap/";QStringList fileNames;THelper::File::BrowseDir(fileNames, colorMapPath, "*.*");if (!COLORMAP_TABLE) {if(fileNames.count() > 0) {COLORMAP_TABLE = new cv::Vec3b[256 * fileNames.count()];BYTE * pDst = (BYTE*)COLORMAP_TABLE;foreach(QString fileName, fileNames) {cv::Mat mat = CvHelper::MatFromFile(fileName);if(mat.cols * mat.channels() == 256 * 3) {memcpy(pDst, mat.data, 256 * 3);pDst += 256 * 3;}}}}
最终的颜色变幻处理逻辑:
// 其它-颜色变幻int paramIndex = 0;FMapIndex = GetParamValue_Int(paramIndex++); // 0: 变幻颜色Mat tempMat;cvtColor(dstMat, tempMat, COLOR_BGR2GRAY);dstMat = cv::Mat(tempMat.rows, tempMat.cols, CV_8UC3);cv::Vec3b * table = COLORMAP_TABLE + 256 * FMapIndex;BYTE * pTemp = tempMat.data;cv::Vec3b * pDst = (cv::Vec3b*)dstMat.data;for (int row = 0; row < tempMat.rows; ++row)for (int col = 0; col < tempMat.cols; ++col)* pDst++ = table[*pTemp++];
本质上还是逐像素处理,这里是查表法,核心逻辑就是* pDst++ = table[*pTemp++];
这个整明白后,那就可以自己设计一个图片作为查表输入,当然有一点点的美工基础更好。
API函数
OpenCV中,提供了一个applyColorMap函数,具体文档说明:
/** @brief Applies a GNU Octave/MATLAB equivalent colormap on a given image.@param src The source image, grayscale or colored of type CV_8UC1 or CV_8UC3.
@param dst The result is the colormapped source image. Note: Mat::create is called on dst.
@param colormap The colormap to apply, see #ColormapTypes
*/
CV_EXPORTS_W void applyColorMap(InputArray src, OutputArray dst, int colormap);
继续查看ColormapTypes定义,可知其系统提供了22种颜色风格。
//! GNU Octave/MATLAB equivalent colormaps
enum ColormapTypes
{COLORMAP_AUTUMN = 0, //!< ![autumn](pics/colormaps/colorscale_autumn.jpg)COLORMAP_BONE = 1, //!< ![bone](pics/colormaps/colorscale_bone.jpg)COLORMAP_JET = 2, //!< ![jet](pics/colormaps/colorscale_jet.jpg)COLORMAP_WINTER = 3, //!< ![winter](pics/colormaps/colorscale_winter.jpg)COLORMAP_RAINBOW = 4, //!< ![rainbow](pics/colormaps/colorscale_rainbow.jpg)COLORMAP_OCEAN = 5, //!< ![ocean](pics/colormaps/colorscale_ocean.jpg)COLORMAP_SUMMER = 6, //!< ![summer](pics/colormaps/colorscale_summer.jpg)COLORMAP_SPRING = 7, //!< ![spring](pics/colormaps/colorscale_spring.jpg)COLORMAP_COOL = 8, //!< ![cool](pics/colormaps/colorscale_cool.jpg)COLORMAP_HSV = 9, //!< ![HSV](pics/colormaps/colorscale_hsv.jpg)COLORMAP_PINK = 10, //!< ![pink](pics/colormaps/colorscale_pink.jpg)COLORMAP_HOT = 11, //!< ![hot](pics/colormaps/colorscale_hot.jpg)COLORMAP_PARULA = 12, //!< ![parula](pics/colormaps/colorscale_parula.jpg)COLORMAP_MAGMA = 13, //!< ![magma](pics/colormaps/colorscale_magma.jpg)COLORMAP_INFERNO = 14, //!< ![inferno](pics/colormaps/colorscale_inferno.jpg)COLORMAP_PLASMA = 15, //!< ![plasma](pics/colormaps/colorscale_plasma.jpg)COLORMAP_VIRIDIS = 16, //!< ![viridis](pics/colormaps/colorscale_viridis.jpg)COLORMAP_CIVIDIS = 17, //!< ![cividis](pics/colormaps/colorscale_cividis.jpg)COLORMAP_TWILIGHT = 18, //!< ![twilight](pics/colormaps/colorscale_twilight.jpg)COLORMAP_TWILIGHT_SHIFTED = 19, //!< ![twilight shifted](pics/colormaps/colorscale_twilight_shifted.jpg)COLORMAP_TURBO = 20, //!< ![turbo](pics/colormaps/colorscale_turbo.jpg)COLORMAP_DEEPGREEN = 21 //!< ![deepgreen](pics/colormaps/colorscale_deepgreen.jpg)
};
运行效果
OpenCV 4 功能 - 颜色变幻
不过,用applyColorMap的话,受到的制约就是只有这22种效果。也许OpenCV 5会更多一些。要想自由的话,还是直接查表处理,那里啥都有。