## computer vision – OpenCV Decompose projection matrix

I got confused with the outputs of opencv decomposeProjectionMatrix function.

I have the projection matrix and the camera matrix “K” and I want to get the translation vector “t”and the Rotation matrix “R” from the projection matrix

As I know the projection matrix of dimension 34 = K[R|t] in which “t” is a 31 vector

cv2.decomposeProjectionMatrix returns R with dimension 33 which is correct but the transVect returned is of dimension 41 not 3*1

My question is how to get back the projection matrix from the function outputs?

documentation: https://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html

## computer vision – Matching superimposed image

We are given two grayscale images, one of which contains a large, mostly contiguous patch from the other one. The patch can be altered with noise, its levels may be stretched, etc.

Here’s an example  We would like to determine the region of the image which was copied onto the other image.

My first instinct was to look at the local correlation. I first apply a little bit of blur to eliminate some of the noise. Then, around each point, I can subtract a gaussian average, then look at the covariance weighted by that same Gaussian kernel. I normalize by the variances, measured in the same way, to get a correlation. If $$G$$ is the Gaussian blur operator, this is:

$$frac{G(A times B) – G(A)G(B)}{sqrt{(G(A^2)-G(A)^2)(G(B^2)-G(B)^2)}$$

The result is… not too bad, not great: Playing with the width of the kernel can help a bit. I’ve also try correlating Laplacians instead of the images themselves, but it seems to hurt more than it helps. I’ve also tried using the watershed algorithm on the correlation, and it just didn’t give very good results.

I’m thinking part of my problem is not having a strong enough prior for what the patch should be like, perhaps a MRF would help here? Besides MRF, are there some other techniques, perhaps more lightweight that would apply? The other part is that correlation doesn’t seem to be all that great at measuring the distance. There are places where the correlation is very high despite the images being very visually distinct. What other metrics could be of use?

## amazon web services – Practice assignment AWS Computer Vision : get_Cifar10_dataset

I have probleme with this methode which should return both the training and the validation dataset and examine it to return the index that corresponds to the first occurrence of each class in CIFAR10.
this is code :
def get_cifar10_dataset():
“””
Should create the cifar 10 network and identify the dataset index of the first time each new class
appears

:return: tuple of training and validation dataset as well as label indices
:rtype: (gluon.data.Dataset, 'dict_values' object is not subscriptable, gluon.data.Dataset,
dict(int:int))
"""

train_data = None
val_data = None
train_data = datasets.CIFAR10(train=True, root=M5_IMAGES)
val_data = datasets.CIFAR10(train=False, root=M5_IMAGES)


## computer vision – Are there some papers or books that compactly elaborate the essential part of the multiple view geometry?

I am a novice to visual SLAM and computer vision. Recently I’ve read a paper “A micro Lie theory for state estimation in robotics” dedicated to introducing the essential part that necessary to roboticists.

I am wondering that are there some papers or books that compactly elaborate the essential part of the multiple view geometry in the manner like the former paper?

And as for bundle adjustment, any recommended papers?

## computer vision: 3D reconstruction from multiple images

Sorry if it is a trivial question but I am a beginner.

I am working on the 3D reconstruction of the multi-image project.

I am working with these data sets.

There is a par.txt file that includes the intrinsic and extrinsic parameters of the camera.
As far as I know, the extrinsic parameter contains the translation and rotation of the W.R.T. coordinate. of the camera. The world coordinate.

How do I know where the origin of world coordinates is?

## computer vision – importing poblem for images

Thanks for contributing with a response to Computer Science Stack Exchange!

But avoid

• Make statements based on opinion; back them up with references or personal experience.

Use MathJax to format equations. MathJax reference.

## Leftist Loons, Lemme get clear vision: Americans must stay home until further notice, but can foreigners enter directly?

🥴 Trump closed all international traffic to the US USA

I am 100% for that. And I'm furious that he didn't do this a long time ago.

Yes, we have to stay locked up because South Korea reopened too soon and caught a second wave. THAT'S WHY EXPERTS SAY REOPENING IS COMING SOON.

So what are you talking about?

I'm reading chapter 1 of gonzalez
As highlighted in the attached snapshot, what is meant by this mid-level task:
description
of those objects to reduce them to a form suitable for computer processing,

please kindly elaborate / explain this statement, preferably with an example ## image processing: I do not understand the meaning of the machine vision paper formula

I am a master's student in the Computer Department.
I started the master in this year.
I am now reading two articles on computer vision.
Actually, before time I had no experience in machine vision. Therefore, I am not familiar with the concepts used in this field.

OK. I want to start question.
1) Template matching using fast normalized cross correlation
by "Kai Briechle and Uwe D. Hanebeck
2) Quick template matching
by J.P. Lewis

Question 1. In the "2) document above, we can see the formula as below, f (x, y) is the image source, t (x, y) is
Template image. I wonder why two terms in the red box are constant and eliminate. I appreciate that only mathematically.
Who can explain me? using a photo or easily. I do not understand by field of image processing.

Question 2. In the "1) previous document" I do not understand that the red box term (average value) is that.
because, in general, we can calculate the average value in this way (value1 + value2 / 2) ## focus: let's see if my vision of the moon illusion is correct

Let's see if my vision of the moon illusion is correct: The true cause of the moon illusion?

As shown in the figure, the blue line is the lens, w is the height of the object, x is the height of the image, v is the distance of the image, u is the distance of the object, and f is the focal distance. The red line is the path of light.
The relationship between u, v, f is

1 / u + 1 / v = 1 / f

and so

f = uv / (v + u) (1)

The observer's eyes are constant, so v is fixed. The distance between the observer and the object is constant, so u is also fixed. Once v and u are fixed, we can tell from equation (1) that f is also fixed. If v is fixed and u decreases, then f also decreases.

x / w = (v-f) / f = v / f-1

and so

x = w (v / f-1) (2)

According to formula (2), if v and w are fixed, x will increase when f decreases.

When the observer observes the moon on the horizon, due to the influence of mountains and trees, f is smaller than that when observing the moon at zenith. According to formula (2), we can know that when f decreases, x increases. Then the observer will feel that the moon on the horizon is larger and closer than the moon at zenith.

I think this is the reason for the moon illusion.

Simple calculation

When looking at nearby trees with your eyes:

u = 200 m (assuming 200 m from the tree)

v = 0.024 m (diameter of the eyeball, assumed length of the image)

w = 10 m (assuming the tree is 10 m high)

f = uv / (v + u)
= 0.0239971 m

x = w (v / f-1)
= 0.0012m = 1.2mm (height of tree image)

When looking at the zenith moon (without the influence of the trees on the ground) with the eyes:

u = 380000000 m (distance from the observer to the moon)

v = 0.024 m

w = 3476000 meters (moon diameter)

f = uv / (v + u) = A (we set this focal length to A)

x = w (v / f-1)
= 0.000219537 m = 0.219537 mm

In the direction of the horizon, if you look at the moon at the focal length of the observation tree:

f = 0.0239971 m

x = w (v / f-1)
= 420,067 m

Observing the image of the moon at the zenith is 0.219537 mm, and observing the image of the moon on the horizon is 420.067 m, showing a great difference between the two. So using a focal length less than A will "magnify" the moon.

Of course, the eyes generally do not observe the moon with a focal length of 0.0239971 m. Because the image may not be clear. But if the moon image is not clear at this focal length, then the eyes will adjust the focal length. Adjust to a focal length that is clear to the image. This focal length is less than A, but it is the focal length for clear images. Because the moon is so far away, the depth of field of the moon image is very large. Therefore, there is a focal length that is smaller than A and can display images clearly. So the moon illusion is caused by a relatively short focal length. I think that's why the illusion of the moon.

reference

https://en.wikipedia.org/wiki/Moon_illusion