106~109 features and descriptor

Author: JimmyHHua

Diff for: README.md

@@ -143,4 +143,8 @@ code_101 | [HOG Multiscale Detection](python/code_101/opencv_101.py)   | ✏️
 code_102 | [HOG Extract Descriptor](python/code_102/opencv_102.py)   | ✔️
 code_103 | [HOG Use Descriptors to Generate Sample Data](python/code_103/opencv_103.py)   | ✔️
 code_104 | [(Detection Case)-HOG+SVM Train](python/code_104/opencv_104.py)   | ✔️
-code_105 | [(Detection Case)-HOG+SVM Predict](python/code_105/opencv_105.py)   | ✔️
+code_105 | [(Detection Case)-HOG+SVM Predict](python/code_105/opencv_105.py)   | ✔️
+code_106 | [AKAZE Features and Descriptors](python/code_106)   | ✔️
+code_107 | [Brisk Features and Descriptors](python/code_107)   | ✔️
+code_108 | [GFTT Detector](python/code_108)   | ✔️
+code_109 | [BLOB Feature Analysis](python/code_109)   | ✔️

Diff for: README_CN.md

@@ -142,4 +142,8 @@ code_101 | [HOG 多尺度检测](python/code_101/opencv_101.py)   | ✏️
 code_102 | [HOG 提取描述子](python/code_102/opencv_102.py)   | ✔️
 code_103 | [HOG 使用描述子生成样本数据](python/code_103/opencv_103.py)   | ✔️
 code_104 | [(检测案例)-HOG+SVM 训练](python/code_104/opencv_104.py)   | ✔️
-code_105 | [(检测案例)-HOG+SVM 预测](python/code_105/opencv_105.py)   | ✔️
+code_105 | [(检测案例)-HOG+SVM 预测](python/code_105/opencv_105.py)   | ✔️
+code_106 | [AKAZE 特征与描述子](python/code_106)   | ✔️
+code_107 | [Brisk 特征与描述子](python/code_107)   | ✔️
+code_108 | [GFTT关键点检测](python/code_108)   | ✔️
+code_109 | [BLOB 特征分析](python/code_109)   | ✔️

Diff for: python/code_106/README.md

@@ -0,0 +1,34 @@
+### AKAZE特征与描述子
+
+AKAZE特征提取算法是局部特征描述子算法，可以看成是SIFT算法的改进、采用非线性扩散滤波迭代来提取与构建尺度空间、采用与SIFT类似的方法寻找特征点、在描述子生成阶段采用ORB类似的方法生成描述子，但是描述子比ORB多了旋转不变性特征。
+
+ORB采用LDB方法，AKAZE采用 M-LDB。
+
+**代码：**
+
+```python
+import cv2 as cv
+
+box = cv.imread("box.png");
+box_in_sence = cv.imread("box_in_scene.png");
+cv.imshow("box", box)
+cv.imshow("box_in_sence", box_in_sence)
+
+# 创建AKAZE特征检测器
+akaze = cv.AKAZE_create()
+kp1, des1 = akaze.detectAndCompute(box,None)
+kp2, des2 = akaze.detectAndCompute(box_in_sence,None)
+
+# 暴力匹配
+bf = cv.BFMatcher(cv.NORM_HAMMING, crossCheck=True)
+matches = bf.match(des1,des2)
+
+# 绘制匹配
+result = cv.drawMatches(box, kp1, box_in_sence, kp2, matches, None)
+cv.imshow("orb-match", result)
+cv.imwrite('orb-match.jpg', result)
+cv.waitKey(0)
+cv.destroyAllWindows()
+```
+
+![](orb-match.jpg)

Diff for: python/code_106/box.png

@@ -0,0 +1,24 @@
+import cv2 as cv
+
+box = cv.imread("box.png");
+box_in_sence = cv.imread("box_in_scene.png");
+cv.imshow("box", box)
+cv.imshow("box_in_sence", box_in_sence)
+
+# 创建AKAZE特征检测器
+akaze = cv.AKAZE_create()
+kp1, des1 = akaze.detectAndCompute(box,None)
+kp2, des2 = akaze.detectAndCompute(box_in_sence,None)
+
+# 暴力匹配
+bf = cv.BFMatcher(cv.NORM_HAMMING, crossCheck=True)
+matches = bf.match(des1,des2)
+
+# 绘制匹配
+result = cv.drawMatches(box, kp1, box_in_sence, kp2, matches, None)
+cv.imshow("orb-match", result)
+cv.imwrite('orb-match.jpg', result)
+cv.waitKey(0)
+cv.destroyAllWindows()
+
+

Diff for: python/code_106/box_in_scene.png

@@ -0,0 +1,34 @@
+### Brisk特征提取与描述子匹配
+
+BRISK(Binary robust invariant scalable keypoints)是一种基于尺度空间不变性类似ORB特征描述子的特征提取算法。BRISK主要步骤可以分为如下两步：
+1.	构建尺度空间金字塔实现关键点定位
+2.	根据关键点生成描述子
+
+**代码：**
+
+```python
+import cv2 as cv
+
+box = cv.imread("box.png");
+box_in_sence = cv.imread("box_in_scene.png");
+cv.imshow("box", box)
+cv.imshow("box_in_sence", box_in_sence)
+
+# 创建BRISK特征检测器
+brisk = cv.BRISK_create()
+kp1, des1 = brisk.detectAndCompute(box,None)
+kp2, des2 = brisk.detectAndCompute(box_in_sence,None)
+
+# 暴力匹配
+bf = cv.BFMatcher(cv.NORM_HAMMING, crossCheck=True)
+matches = bf.match(des1,des2)
+
+# 绘制匹配
+result = cv.drawMatches(box, kp1, box_in_sence, kp2, matches, None)
+cv.imshow("orb-match", result)
+cv.imwrite('orb-match.jpg', result)
+cv.waitKey(0)
+cv.destroyAllWindows()
+```
+
+![](orb-match.jpg)

Diff for: python/code_106/opencv_106.py

@@ -0,0 +1,22 @@
+import cv2 as cv
+
+box = cv.imread("box.png");
+box_in_sence = cv.imread("box_in_scene.png");
+cv.imshow("box", box)
+cv.imshow("box_in_sence", box_in_sence)
+
+# 创建BRISK特征检测器
+brisk = cv.BRISK_create()
+kp1, des1 = brisk.detectAndCompute(box,None)
+kp2, des2 = brisk.detectAndCompute(box_in_sence,None)
+
+# 暴力匹配
+bf = cv.BFMatcher(cv.NORM_HAMMING, crossCheck=True)
+matches = bf.match(des1,des2)
+
+# 绘制匹配
+result = cv.drawMatches(box, kp1, box_in_sence, kp2, matches, None)
+cv.imshow("orb-match", result)
+cv.imwrite('orb-match.jpg', result)
+cv.waitKey(0)
+cv.destroyAllWindows()

Diff for: python/code_106/orb-match.jpg

@@ -0,0 +1,39 @@
+### 特征提取之关键点检测 - GFTTDetector
+
+该方法是基于shi-tomas角点检测变化而来的一种特征提取方法，OpenCV创建该检测器的API与goodfeaturetotrack的API参数极其类似：
+```
+Ptr<GFTTDetector> cv::GFTTDetector::create(
+	int maxCorners = 1000,
+	double qualityLevel = 0.01,
+	double minDistance = 1,
+	int blockSize = 3,
+	bool useHarrisDetector = false,
+	double k = 0.04
+)
+```
+唯一不同的，该方法返回一个指针。
+
+PS:
+>需要注意的是该方法无法提取描述子，只支持提取关键点！
+
+**代码：**
+
+```python
+import cv2 as cv
+
+
+image = cv.imread("hist_02.jpg");
+cv.imshow("input", image)
+
+# 创建GFTT特征检测器
+gftt = cv.GFTTDetector_create(100, 0.01,1, 3, False, 0.04)
+kp1 = gftt.detect(image,None)
+result = cv.drawKeypoints(image, kp1, None, (0, 255, 0), cv.DrawMatchesFlags_DEFAULT)
+
+cv.imshow("GFTT-Keypoint-Detect", result)
+cv.imwrite("GFTT-Keypoint-Detect.jpg", result)
+cv.waitKey(0)
+cv.destroyAllWindows()
+```
+
+![](GFTT-Keypoint-Detect.jpg)

Diff for: python/code_107/README.md

@@ -0,0 +1,15 @@
+import cv2 as cv
+
+
+image = cv.imread("hist_02.jpg");
+cv.imshow("input", image)
+
+# 创建GFTT特征检测器
+gftt = cv.GFTTDetector_create(100, 0.01,1, 3, False, 0.04)
+kp1 = gftt.detect(image,None)
+result = cv.drawKeypoints(image, kp1, None, (0, 255, 0), cv.DrawMatchesFlags_DEFAULT)
+
+cv.imshow("GFTT-Keypoint-Detect", result)
+cv.imwrite("GFTT-Keypoint-Detect.jpg", result)
+cv.waitKey(0)
+cv.destroyAllWindows()

Diff for: python/code_107/box.png

@@ -0,0 +1,72 @@
+### BLOB特征分析 – simpleblobdetector使用
+
+BLOB是图像中灰度块的一种专业称呼，更加变通一点的可以说它跟我们前面二值图像分析的联通组件类似，通过特征提取中的SimpleBlobDetector可以实现常见的各种灰度BLOB对象组件检测与分离。使用该检测器的时候，可以根据需要输入不同参数，得到的结果跟输入的参数息息相关。常见的BLOB分析支持如下：
+-   根据BLOB面积过滤
+-   根据灰度/颜色值过滤
+-   根据圆度过滤
+-   根据长轴与短轴过滤
+-   根据凹凸进行过滤
+
+**参数列表:**
+```
+SimpleBlobDetector::Params::Params()
+bool    filterByArea
+bool    filterByCircularity
+bool    filterByColor
+bool    filterByConvexity
+bool    filterByInertia
+float   maxArea
+float   maxCircularity
+float   maxConvexity
+float   maxInertiaRatio
+float   maxThreshold
+float   minArea
+float   minCircularity
+float   minConvexity
+float   minDistBetweenBlobs
+float   minInertiaRatio
+```
+
+代码：
+```python
+import cv2 as cv
+import numpy as np
+
+# load image
+frame = cv.imread("zhifang_ball.png")
+gray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
+params = cv.SimpleBlobDetector_Params()
+
+# Change thresholds
+params.minThreshold = 0;
+params.maxThreshold = 256;
+
+# Filter by Area.
+params.filterByArea = True
+params.minArea = 100
+
+# Filter by Circularity
+params.filterByCircularity = True
+params.minCircularity = 0.1
+
+# Filter by Convexity
+params.filterByConvexity = True
+params.minConvexity = 0.5
+
+# Filter by Inertia
+params.filterByInertia = True
+params.minInertiaRatio = 0.5
+
+detector = cv.SimpleBlobDetector_create(params)
+
+# Detect blobs.
+cv.imshow("input", frame)
+keypoints = detector.detect(gray)
+result = cv.drawKeypoints(frame, keypoints, None, (0, 0, 255), cv.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
+cv.imshow("result", result)
+cv.imwrite("result.png", result)
+cv.waitKey(0)
+cv.destroyAllWindows()
+```
+
+![](result.png)

Diff for: python/code_107/box_in_scene.png

@@ -0,0 +1,38 @@
+import cv2 as cv
+import numpy as np
+
+# load image
+frame = cv.imread("zhifang_ball.png")
+gray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
+params = cv.SimpleBlobDetector_Params()
+
+# Change thresholds
+params.minThreshold = 0;
+params.maxThreshold = 256;
+
+# Filter by Area.
+params.filterByArea = True
+params.minArea = 100
+
+# Filter by Circularity
+params.filterByCircularity = True
+params.minCircularity = 0.1
+
+# Filter by Convexity
+params.filterByConvexity = True
+params.minConvexity = 0.5
+
+# Filter by Inertia
+params.filterByInertia = True
+params.minInertiaRatio = 0.5
+
+detector = cv.SimpleBlobDetector_create(params)
+
+# Detect blobs.
+cv.imshow("input", frame)
+keypoints = detector.detect(gray)
+result = cv.drawKeypoints(frame, keypoints, None, (0, 0, 255), cv.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
+cv.imshow("result", result)
+cv.imwrite("result.png", result)
+cv.waitKey(0)
+cv.destroyAllWindows()