Problem Description:
After using the Deeplab V3 + divided pictures obtained, the next step hope for the classes and the proportion of the pixel area, so that other aspects of the analysis, the extraction process is as follows:
Use OpenCV extraction area:
Main process: 1 to the image from BGR HSV, color-coded to facilitate;
2. Use cv2.inRange function, extracts a specified color, such as: red;
3. Find a profile, and calculate the area profile.
The whole process made animation, is Jiang Zi:
Difficulties: How to find HSV values of the image of it?
Method 1: Use the OpenCV cv2.cvtColor conversion, HSV values obtained can then set the lower limit. however,
Some colors RGB eye can not see, the second method should be adopted at this time.
red = np.uint8([[[0, 0, 128]]])
hsv_red = cv2.cvtColor(red, cv2.COLOR_BGR2HSV)
print(hsv_red) # [[[ 0 255 128]]]
Method 2: Using PS HSV values obtained. Click on figure 1 (icon like straw), select HSB color, the figure at 2 shows the results.
Since OpenCV, H is [0, 180], S [0, 255], V [0, 255],
H is in PS [0, 360], S [0, 1], V [0, 1],
Therefore, conversion: H data in PS / 2, S x 255, V x 255 HSV values obtained in OpenCV.