jjbluesky's profile - activity

I agree it looks like that, but all the mats have exactly the same size as the original image (1536 , 2048). So I don't think the image size changes somewhere in the process.

Hi,

I've implemented the following threshold algorithm in OpenCV4Android:

1) Import image: inputImg = Highgui.imread("photo.jpg");

2) Grayscale: Imgproc.cvtColor(inputImg, test3, Imgproc.COLOR_BGRA2GRAY, 1);

3) Blur: Imgproc.blur(test3, test2, new Size(83, 83));

4) Substract: Core.subtract(test2, test3, test4);

5) Threshold: Imgproc.threshold(test4, test3, 50, 255, Imgproc.THRESH_BINARY_INV);

6) RGBA: Imgproc.cvtColor(test3, inputImg, Imgproc.COLOR_GRAY2BGRA, 4);

It works fine, but when I zoom into the image I get these "steps" (pixel artifacts) on the edges of the letters. So I tried the same algorithm in Matlab and "regular" OpenCV....and the results are much better. That confuses me and I have no idea what I could do to improve the result in OpenCV4Android.

Top left = OpenCV4Android

Top right = Original image

Bottom = Matlab / OpenCV (looks pretty much the same)

Does anyone have a clue why the results are so different or what I could do to improve the OpenCV4Android image?

Thanks, Jana

True story. It finally works. I'm happy :)

I have the same problem. At this point...I just feel stupid. I did everything exactly like in the tutorial...I've read all the related questions in the forum and I'm still getting a black screen. I'm using OpenCV4Android 2.4.3 & Android Version 4.1.2. I'm also using an OpenCV4Android example without JNI....so it's not that difficult...actually. But I just can't figure it out -.- ...

Oh wow! I totally missed the update. That's a cool new tutorial. Thanks for telling me! Unfortunately I can't set the focus mode to continuous picture.

Which one is tutorial 5? I couldn't find any with direct camera control. The only ones I have are here: http://opencv.org/platforms/android/opencv4android-samples.html

Can you tell me where I can set the focus mode? I'm using the opencv4android sample projects as a basis ...and there is no direct camera access. Everything works via CameraBridgeViewBase.

Well I waited patiently for the focus...but it's not only the focus. There is a big different....I attached two pictures as an example. Just look at the font. I can't imagine how this is happening...the camera has 8 mpx.

Thanks, you gave me the push in the right direction :).

Hi,

maybe it's a silly question...anyway:

I compared the image/video quality of the preview image of the camera app of an android phone (samsung galaxy s3) and the example apps which come with the opencv4android framework.

So why is it that the preview of the camera app is so much better then the preview of the opencv4android apps? I hope it's visible on the screenshots --> the fonts in the camera app are a lot sharper than in the opencv4android app.

Thanks :)!

I attached 2 pictures for a better understanding. The font in the camera app screenshot is so much clearer than the opencv4android screenshot.

Is there a way to change it?

camera app screenshot:

opencv4android screenshot:

I'm using version 2.4.3. I don't know how to process the stopped camera image....if that's possible.

Hi,

I have an application built on the basis of the sample4mixed example from OpenCV4Android. What I want to do is: I want to "freeze" the camera input in the onCameraFrame() method. Then I want to zoom into the frozen picture (e.g. 4x) and drag the zoomed part around on the entire frozen picture.

But I have no idea so far how to do that. I don't know how to "freeze" the camera input. I know how to zoom into a video, so maybe that's the same with the picture then...but I have not a clue how to drag the picture around...if that makes sense. If not let me know and I try to explain it a little bit more.

P.S. The image should not be saved on SD card or anything.

Thanks :)!

Hi,

I was wondering which might be the best method (performance-wise) to sharpen real-time video frames (I am using an Android phone for this).

I have: 2x/4x/8x (digitally) zoomed real-time video frames

My problem: 4x/8x zoom is pretty blurry

I want to: sharpen the zoomed video frames in real-time

One idea is to use a laplacian filter on the zoomed video frames. The question is: is this the best idea performance-wise? What else could I use? Maybe I should smooth the video frames before I use the laplacian filter..

Any thoughts on the topic are appreciated :)!

This is how my images look now (screenshots from a phone):

Regular video frame with zoom:

Video frame + inverted canny edge with zoom: