I have to complete the project in which I have to apply some techniques or model to images . My aim is to convert the image to frontal/straight when it comes to skewed , sheared or any transformation. I can use open-cv as well. If there is any tutorial / code , share the link with me ?
-
$\begingroup$ Do you know what type of transformations applied to the images to become like this? $\endgroup$– user119783Jun 29, 2021 at 7:29
-
$\begingroup$ Any transformation means sheared , skewed , rotation etc .. $\endgroup$– HamzaJun 29, 2021 at 7:31
-
$\begingroup$ So, you have a transformed image without knowing which transformations applied, this is the case, right? It does not matter the type, what matters is your knowledge about the transformations $\endgroup$– user119783Jun 29, 2021 at 7:43
-
$\begingroup$ Yes , exactly , I have no concerns about what type of transformations is applied , the goal i have is just to convert the image to frontal face , whether there is a depth , rotation , skewness , tilted etc.. $\endgroup$– HamzaJun 29, 2021 at 7:45
1 Answer
I suggest understanding first what types of transformations applied to your images and based on that you can select the right reverse transformation to your images because there are different types of transformations e.g., geometric, color transformation, and I will give you an example below:
If the transformations are only geometric transformations, not color transformation, you can read more about perspective transformations see this link to understand more perspective transform in opencv
To retrieve the original image from the transformed image automatically you should use some techniques such as edge detection and finding contours, you can read this blog on how to do that: 1 and also this 2.
Hope this helps!
-
$\begingroup$ I got your point , and I have the use case of shearness not all type of transformations , If image is sheared then how do i unsheared it ? $\endgroup$– HamzaJun 29, 2021 at 8:21
-
$\begingroup$ Your images are transformed by an affine transformation. Then, the answer is above, you have either perspective transform or apply some other techniques prior to make the process more automatic. $\endgroup$– user119783Jun 29, 2021 at 8:35