liuruoze · loukey · Aug 1, 2017
diff --git a/easyPR-doc/algorithm.md b/easyPR-doc/algorithm.md
@@ -0,0 +1,4 @@
+ - NMS(非极大值抑制)
+   - 选取邻近区域中得分最高的矩形块,以区分多个矩形重叠的情况
+ - SVM(支持向量机)
+   - <img src="./img/SVM.png" wid=600px/>
diff --git a/easyPR-doc/classDemo.md b/easyPR-doc/classDemo.md
@@ -0,0 +1,9 @@
+### Class
+
+#### Properties
+
+#### Functions
+
+***
+
+***
diff --git a/easyPR-doc/core/chars_identify.md b/easyPR-doc/core/chars_identify.md
@@ -0,0 +1,34 @@
+### Class CharsIdentify
+
+#### Properties
+ - (private:)
+   - [`annCallback extractFeature`](#extractFeature)
+   - [`static CharsIdentify* instance_`](#instance_)
+   - [`cv::Ptr<cv::ml::ANN_MLP> ann_`](#ann_)
+   - [`cv::Ptr<cv::ml::ANN_MLP> annChinese_`](#annChinese_)
+   - [`cv::Ptr<cv::ml::ANN_MLP> annGray_`](#annGray_)
+   - [`std::shared_ptr<Kv> kv_`](#kv_)
+
+#### Functions
+ - (public:)
+   - [`static CharsIdentify* instance()`](#instance)
+   - [`int classify(cv::Mat f, float& maxVal, bool isChinses = false, bool isAlphabet = false)`](#classify1)
+   - [`void classify(cv::Mat featureRows, std::vector<int>& out_maxIndexs,std::vector<float>& out_maxVals, std::vector<bool> isChineseVec)`](#classify2)
+   - [`void classify(std::vector<CCharacter>& charVec)`](#classify3)
+   - [`void classifyChinese(std::vector<CCharacter>& charVec)`](#classifyChinese)
+   - [`void classifyChineseGray(std::vector<CCharacter>& charVec)`](#classifyChineseGray)
+   - [`std::pair<std::string, std::string> identify(cv::Mat input, bool isChinese = false, bool isAlphabet = false)`](#identify1)
+   - [`int identify(std::vector<cv::Mat> inputs, std::vector<std::pair<std::string, std::string>>& outputs,std::vector<bool> isChineseVec)`](#identify2)
+   - [`std::pair<std::string, std::string> identifyChinese(cv::Mat input, float& result, bool& isChinese)`](#identifyChinese)
+   - [`std::pair<std::string, std::string> identifyChineseGray(cv::Mat input, float& result, bool& isChinese)`](#identifyChineseGray)
+   - [`bool isCharacter(cv::Mat input, std::string& label, float& maxVal, bool isChinese = false)`](#isCharacter)
+   - [`void LoadModel(std::string path)`](#LoadModel)
+   - [`void LoadChineseModel(std::string path)`](#LoadChineseModel)
+   - [`void LoadGrayChANN(std::string path)`](#LoadGrayChANN)
+   - [`void LoadChineseMapping(std::string path)`](#LoadChineseMapping)
+ - (private:)
+   - [`CharsIdentify()`](#CharsIdentify)
+
+***
+
+***
diff --git a/easyPR-doc/core/feature.md b/easyPR-doc/core/feature.md
@@ -0,0 +1,85 @@
+### feature
+
+#### Functions
+ - [`cv::Mat getHistogram(cv::Mat in)`](#getHistogram)
+ - [`typedef void (*svmCallback)(const cv::Mat& image, cv::Mat& features)`](#svmCallback)
+ - [`typedef void (*annCallback)(const cv::Mat& image, cv::Mat& features)`](#annCallback)
+ - [`void getGrayPlusProject(const cv::Mat& grayChar, cv::Mat& features)`](#getGrayPlusProject)
+ - [`void getHistogramFeatures(const cv::Mat& image, cv::Mat& features)`](#getHistogramFeatures)
+ - [`void getSIFTFeatures(const cv::Mat& image, cv::Mat& features)`](#getSIFTFeatures)
+ - [`void getHOGFeatures(const cv::Mat& image, cv::Mat& features)`](#getHOGFeatures)
+ - [`void getHSVHistFeatures(const cv::Mat& image, cv::Mat& features)`](#getHSVHistFeatures)
+ - [`void getLBPFeatures(const cv::Mat& image, cv::Mat& features)`](#getLBPFeatures)
+ - [`void getColorFeatures(const cv::Mat& src, cv::Mat& features)`](#getColorFeatures)
+ - [`void getHistomPlusColoFeatures(const cv::Mat& image, cv::Mat& features)`](#getHistomPlusColoFeatures)
+ - [`cv::Mat charFeatures(cv::Mat in, int sizeData)`](#charFeatures)
+ - [`cv::Mat charFeatures2(cv::Mat in, int sizeData)`](#charFeatures2)
+ - [`void getLBPplusHistFeatures(const cv::Mat& image, cv::Mat& features)`](#getLBPplusHistFeatures)
+ - [`void getGrayCharFeatures(const cv::Mat& grayChar, cv::Mat& features)`](#getGrayCharFeatures)
+ - [`void getGrayPlusLBP(const Mat& grayChar, Mat& features)`](#getGrayPlusLBP)
+
+***
+<a id="getHistogram"></a>
+#### cv::Mat getHistogram(cv::Mat in)
+ - 获得车牌的特征数
+
+<a id="svmCallback"></a>
+#### typedef void (*svmCallback)(const cv::Mat& image, cv::Mat& features)
+ - EasyPR的getFeatures回调函数
+ - 用于从车牌的image生成svm的训练特征features
+
+<a id="annCallback"></a>
+#### typedef void (*annCallback)(const cv::Mat& image, cv::Mat& features)
+ - EasyPR的getFeatures回调函数
+ - convert from images to features used by gray char ann
+
+<a id="getGrayPlusProject"></a>
+#### void getGrayPlusProject(const cv::Mat& grayChar, cv::Mat& features)
+ - gray and project feature
+
+<a id="getHistogramFeatures"></a>
+#### void getHistogramFeatures(const cv::Mat& image, cv::Mat& features)
+ - EasyPR的getFeatures回调函数
+ - 本函数是获取垂直和水平的直方图图值
+
+<a id="getSIFTFeatures"></a>
+#### void getSIFTFeatures(const cv::Mat& image, cv::Mat& features)
+ - 本函数是获取SIFT特征子
+
+<a id="getHOGFeatures"></a>
+#### void getHOGFeatures(const cv::Mat& image, cv::Mat& features)
+ - 本函数是获取HOG特征子
+
+<a id="getHSVHistFeatures"></a>
+#### void getHSVHistFeatures(const cv::Mat& image, cv::Mat& features)
+ - 本函数是获取HSV空间量化的直方图特征子
+
+<a id="getLBPFeatures"></a>
+#### void getLBPFeatures(const cv::Mat& image, cv::Mat& features)
+ - LBP feature
+
+<a id="getColorFeatures"></a>
+#### void getColorFeatures(const cv::Mat& src, cv::Mat& features)
+ - color feature
+
+<a id="getHistomPlusColoFeatures"></a>
+#### void getHistomPlusColoFeatures(const cv::Mat& image, cv::Mat& features)
+ - color feature and histom
+
+<a id="charFeatures"></a>
+#### cv::Mat charFeatures(cv::Mat in, int sizeData)
+ - get character feature
+
+<a id="charFeatures2"></a>
+#### cv::Mat charFeatures2(cv::Mat in, int sizeData)
+ - get character feature
+
+<a id="getLBPplusHistFeatures"></a>
+#### void getLBPplusHistFeatures(const cv::Mat& image, cv::Mat& features)
+ - LBP feature + Histom feature
+
+<a id="getGrayCharFeatures"></a>
+#### void getGrayCharFeatures(const cv::Mat& grayChar, cv::Mat& features)
+
+<a id="getGrayPlusLBP"></a>
+#### void getGrayPlusLBP(const Mat& grayChar, Mat& features)
diff --git a/easyPR-doc/core/plate.md b/easyPR-doc/core/plate.md
@@ -0,0 +1,16 @@
+### Class
+
+#### Properties
+ - (private:)
+   - [`Mat m_plateMat`](#m_plateMat)
+   - [`RotatedRect m_platePos`](#m_platePos)
+   - [`double m_score`](#m_score)
+
+#### Functions
+
+***
+<a id="m_score"></a>
+####double m_score
+ - 保存了`SVM`后的分数,越大证明越可能是车牌
+
+***
diff --git a/easyPR-doc/core/plate_detect.md b/easyPR-doc/core/plate_detect.md
@@ -0,0 +1,29 @@
+### Class CPlateDetect
+
+#### Properties
+ - (priavte:)
+   - [`int m_maxPlates`](#m_maxPlates)
+   - [`CPlateLocate* m_plateLocate`](#m_plateLocate)
+   - [`int m_type`](#m_type)
+   - [`static std::string m_pathSvm`](#m_pathSvm)
+   - [`bool m_showDetect`](#m_showDetect)
+
+#### Functions
+ - (public:)
+   - [`int plateDetect(Mat src, s td::vector<CPlate> &resultVec, int type,bool showDetectArea, int img_index = 0)`](#plateDetect1)
+   - [`int plateDetect(Mat src, std::vector<CPlate> &resultVec, int img_index = 0)`](#plateDetect2)
+   - [`void LoadSVM(std::string s)`](#LoadSVM)
+
+***
+<a id="m_maxPlates"></a>
+#### int m_maxPlates(`get|set`)
+
+<a id="m_plateLocate"></a>
+#### CPlateLocate* m_plateLocate
+ - 用于车牌定位的对象,完成了所有车牌定位的操作
+
+<a id="m_type"></a>
+#### m_type(`set`)
+ -
+
+***
diff --git a/easyPR-doc/core/plate_judge.md b/easyPR-doc/core/plate_judge.md
@@ -0,0 +1,71 @@
+### Class PlateJudge
+
+#### Properties
+ - (private:)
+   - [`static PlateJudge* instance_`](#instance_)
+   - [`svmCallback extractFeature`](#extractFeature)
+   - [`cv::Ptr<ml::SVM> svm_`](#svm_)
+
+#### Functions
+ - (public:)
+   - [`static PlateJudge* instance()`](#instance)
+   - [`void LoadModel(std::string path)`](#LoadModel)
+   - [`int plateJudgeUsingNMS(const std::vector<CPlate>&, std::vector<CPlate>&, int maxPlates = 5)`](#plateJudgeUsingNMS)
+   - [`int plateSetScore(CPlate& plate)`](#plateSetScore)
+   - [`int plateJudge(const Mat& plateMat)`](#plateJudge1)
+   - [`int plateJudge(const std::vector<Mat> &inVec,std::vector<Mat> &resultVec)`](#plateJudge2)
+   - [`int plateJudge(const std::vector<CPlate> &inVec,std::vector<CPlate> &resultVec)`](#plateJudge3)
+   - [`void NMS(std::vector<CPlate> &inVec, std::vector<CPlate> &resultVec, double overlap)`](#NMS)
+   - [`int PlateJudge::plateJudgeUsingNMS(const std::vector<CPlate> &inVec, std::vector<CPlate> &resultVec, int maxPlates)`](#plateJudgeUsingNMS)
+ - (private:)
+   - [`PlateJudge()`](#PlateJudge)
+
+***
+<a id="instance_"></a>
+#### static PlateJudge* instance_
+ - 静态变量`instance_`,用于单例模式
+
+***
+<a id="instance"></a>
+#### static PlateJudge* instance()
+ - 单例模式初始化方法,当`instance_`为`nullptr`时,初始化一个`PlateJudge`
+ - 返回`instance_`
+
+<a id="PlateJudge"></a>
+#### PlateJudge()
+ - 加载`svm`的模型,在`config.h`中配置模型路径
+ - 可选择是否使用`LBP`模型,因为暂时没有这个模型,所以使用的是`Hist`
+
+<a id="LoadModel"></a>
+#### void LoadModel(std::string path)
+ - 输入参数为路径`path`
+ - 如果`path`与默认`kDefaultSvmPath`路径不同,则加载`path`下的模型,如果相同,那么加载的模型和`kDefaultSvmPath`路径下的相同
+
+<a id="plateSetScore"></a>
+#### int plateSetScore(CPlate& plate)
+ - 这个函数通过`SVM`为输入的`CPlate`计算分数,小于0且越小为车牌的可能性越大
+ - 方法判定`score`大于`0.5`的`CPlate`为不是车牌
+ - 返回值为`0`表示是车牌,返回值为`-1`表示不是车牌
+
+<a id="plateJudge1"></a>
+#### int plateJudge(const Mat& plateMat)
+ - 调用`plateSetScore`得到`0/-1`并返回,`0`代表是车牌,`1`表示不是车牌
+
+<a id="plateJudge2"></a>
+#### int plateJudge(const std::vector<Mat> &inVec,std::vector<Mat> &resultVec)
+ - 输入为`Mat`的集合`inVec`,代表了需要判断的`Mat`的集合
+ - 输出为`resultVec`,代表了所有可能是车牌的集合
+ - `return 0`
+
+<a id="plateJudge3"></a>
+#### int plateJudge(const std::vector<CPlate> &inVec,std::vector<CPlate> &resultVec)
+ - 输入为`CPlate`的集合`inVec`,代表了需要判断的`CPlate`的集合
+ - 输出为`resultVec`,代表了所有可能是车牌的集合
+ - 和`int plateJudge(const std::vector<Mat> &inVec,std::vector<Mat> &resultVec)`不同的是,在这个方法中会进行一次矩形的缩小再判断
+ - `return 0`
+
+<a id="NMS"></a>
+#### void NMS(std::vector<CPlate> &inVec, std::vector<CPlate> &resultVec, double overlap)
+
+<a id="plateJudgeUsingNMS"></a>
+#### int PlateJudge::plateJudgeUsingNMS(const std::vector<CPlate> &inVec, std::vector<CPlate> &resultVec, int maxPlates)