Watershed is not helping in this matter. I have tried that. Maps of India are too bad that they can't even detect a contour in this low-quality image. I am going with the active contours on the Edge Detector Output but active contours are taking a lot of computation time. Any better approach would be really appreciated.

I am working on this picture.

Due to its bad quality, firstly I use histogram equalization after that bilateral blur to preserve the edges, adaptive Canny and edge sharpening kernel and the output is this:-

I need the edges to be closed and gone through morphological operations but the results were not satisfactory and the operation can't be generalized on all the images. How to solve this? Here's the code:- Edge Sharp

I need to sharpen the edge of the following image.

I am doing a 2D convolution with the image for edge sharpening. The kernels are [[1,1,1],[1,-7,1],[1,1,1]] and [[-1,-1,-1],[-1,9,-1],[-1,-1,-1]] Then there's this error:- error: (-213) Unsupported combination of source format (=4), and destination format (=4) in function cv::getLinearFilter

On the other hand, I am using it on the gray image and it's working fine. Thanks for the help.

I am taking this image as my input and applying thresholding, median Blur and Watershed algorithm to extract buildings.

After applying Median blur and thresholding I am getting this result

Median blur helps to remove the text but how to segment out the roads?

shifted = cv2.pyrMeanShiftFiltering(image, 21, 51) gray = cv2.cvtColor(shifted, cv2.COLOR_BGR2GRAY) thresh = cv2.threshold(gray, 0, 255,cv2.THRESH_BINARY | cv2.THRESH_OTSU)

Thanks a lot.

I was going through some tutorial on watershed segmentation process and finding contours.

I am working on satellite images to mask out the rooftop area from houses and buildings.

Applying histogram equalization on the image and then applying the watershed algorithm to mask out the buildings, I am getting too many responses and without applying that some rooftops are missing. Is there any way to optimize the number of counts? Here are some sample images that I have to work on.

I am applying Mean Shift Filtering and after that thresholding and then watershed algorithm.

In case of finding contours, cnts = cv2.findContours(mask.copy(), cv2.RETR_EXTERNAL,cv2.CHAIN_APPROX_SIMPLE)[-2] what does -2 implies? Thanks for help.

What is the difference between cv2.threshold(gray, 0, 255,cv2.THRESH_BINARY | cv2.THRESH_OTSU) and cv2.threshold(gray, 0, 255,cv2.THRESH_BINARY + cv2.THRESH_OTSU) ?

# Otsu's thresholding
gray = cv2.cvtColor(shifted, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255,cv2.THRESH_BINARY | cv2.THRESH_OTSU)**[1]**

In the end, there's [1] written what does it mean? I am trying to use automatic threshold.

Rather playing with the parameters of Canny Edge Detection use adaptive Canny Edge Detector. It will give a far better result. Adaptive Canny works on finding median over the image and uses that value for upper and lower threshold. You can go through this documentation:- Zero Parameter Auto Canny Detection://www.pyimagesearch.com/2015/04/06/zero-parameter-automatic-canny-edge-detection-with-python-and-opencv/) I think this will help you find better edges.

Yeah, I have copied the file from C:\opencv\build\x64\vc14\bin to the root directory of Python27 and added the path to the environment variable too. Please, anything you can point it out more that would be lot helpful.Thanks in advance.

opencv_ffmpeg.dll file is added to the PATH and to the root directory of Python27 too.

Shape from Stereo is fine like using right and left image of picture and then using depth map to plot them all in a single image to recover 3d model. Like during day time if we take a photo from left and right then there are shadows of obstacles on the terrace. When applying shape from stereo does that will create a problem?

Harris and SIFT will not work without shadow removal from the image.I have tried this.

I have been through the different documentation but in all of those shadows are removed from video files. I want to remove the shadow from a single image. cv2.BackgroudSubtractorMOG/GGM is not working but MOG2 is working but on video files only. I want to reconstruct 3d model using 2d points in the image but the shadow points are working like outliers.

I have tried .avi and .mp4 format. No extract error messages are getting. The shell restarts after few seconds. No, the video has no jitter when playing through VLC

Video files can't read although camera is working using VideoCapture. I have added the correct ffmpeg file from opencv directory to C:/Python27 environment variable. What to do now, still the video is not working? Here's my simple code to open video file import numpy as np import cv2

Capture video from file

cap = cv2.VideoCapture('vtest.avi')

while True:

ret, frame = cap.read()

if ret == True:

    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

    cv2.imshow('frame',gray)


    if cv2.waitKey(30) & 0xFF == ord('q'):
        print True
        break

else:
    print False
    break

cap.release() cv2.destroyAllWindows()