Windows imshow sometimes showing gray image

edit

lets look some points of the code ( stitching_detailed.cpp )

    img_warped.convertTo(img_warped_s, CV_16S);
    .
    .
    .
    blender->feed(img_warped_s, mask_warped, corners[img_idx]);
    .
    .
    .
    blender->blend(result, result_mask);

the reason for this problem is type of the image is CV_16S .

The solution is converting image to CV_8U

.
.
blender->blend(result, result_mask);
Mat result8u;
result.convertTo(result8u,CV_8U);
imshow("you can see the image now",result8u);

I've got the same problem with cv2.imshow(). I found a solution by modifying the data type, and imshow() can now read the file :

import numpy as np
import cv2

imgL = cv2.imread('tsukuba_l.png', 0)
imgR = cv2.imread('tsukuba_r.png', 0)

stereo = cv2.createStereoBM(numDisparities=16, blockSize=15)
disparity = stereo.compute(imgL, imgR, cv2.CV_32F)

cv2.namedWindow("disparity_tsukuba", 0)
res = cv2.convertScaleAbs(disparity)
cv2.imshow('disparity_tsukuba', res)

cv2.waitKey(0)

The problem is indeed that imshow() works badly with files other than CV_8U file and can result in wrong image (like all gray images) when given 32F file, which are produced by some functions like StereoBM or other : doc. So we need to convert from a CV_32F to a readable format for imshow()

After having spent a whole day searching, does anybody have a solution for this?

There is no one solution that seems te be working for me. And solutiong is a far reach for the meaning.

I suddenly got the same problem, after manipulating my code. I finally figured that it was because I had removed tha code below, which was copied from an example.

Code necessary to make imshow work for me: if cv2.waitKey(1) & 0xFF == ord('q'): break The ncode itselfe has no effect, I hace to use <ctrl> C to stop the program. But without it, my images turn grey.