Errata

Example 2.6 doesn't work!
1) cv::Mat img = cv::imread( argv[1] ),img2;
What undefined matrix "img2" doing there?
I guess it should be "Mat img2;" in new line.

2) What semicolon doing at the end of main function?

I resolved that problem in this way (berofe namedWindow() function):

Mat img1;
Mat img2;
VideoCapture cap;
cap.open(string(argv[1]));
cap >> img1;

Example 2-7 doesn't work!

1) This is example doesn't work:
cv::Mat img_rgb = cv::imread( argv[1] );

My solution:
Mat img_rgb;
Mat img_gry, img_cny;
VideoCapture cap;
cap.open(string(argv[1]));
cap >> img_rgb;

2) cv::cvtColor( img_rgb, img_gry, cv::BGR2GRAY);
There no "BGR2GRAY", only "CV_BGR2GRAY", maybe here should be "CV_RGB2GRAY".

3)There should be "return 0;" at the end of main function.

That program doesn't work!
1) There are no "logPolar" function, but "cvLogPolar".
2) "cvLogPolar" function underline first two parameters: "no suitable conversion function from "cv::Mat" to "CvArr*" exists".
3) Could not find codec parameters.

The book has many typos and other mistakes.
Here's a short list of some of the errors I noticed so far:

Page 26, paragraph 3, just before the code for example 2-6 It says "To avoid this, we want to first run a high-pass filter over the signal".
This is a mistake - a LOW PASS FILTER is used, not high pass.

In many pages the same mistake appears regarding the syntax for the type of array.
In page 46 it says: "All such types are defined in the library header, and have the form cv::{U8,S16,U16,S32,F32,F64}C{1,2,3}.14 For example, cv::F32C3 would imply a 32-bit floating-point three-channel array."
This is very very wrong in 2 ways.
First - it is not cv::{....} but rather CV_....
The second mistake is that the format has the type and bit depth in reverse order. For instance, F32 should be 32F etc.
If I look into the example It should be CV_32FC3 and NOT cv::F32C3.
This mistake occurs in many many places in the book (in examples 2-7, 2-8) and many other places. Basically this mistake appears in all places were setting the type is part of the code or explanation.

Some of the equations in the pdf aren't vectorized, but rather are pixelated non-readable graphics. For example - on page 46, 2nd, 3rd and 5th rows. This happens in more places as well.

On page 52, second paragraph (green code lines) there's a serious mistake setting the cv::MatConstIterator<> iterator and checking it's value in the while loop.

Instead of "cv::MatConstIterator<cv::Vec3f> it = m.begin();"
It should say "cv::MatConstIterator_<cv::Vec3f> it = m.begin<cv::Vec3f>();"

Instead of "while( it != m.end() ) {"
It should say "while( it != m.end<cv::Vec3f>() ) {"

In addition, the while loop uses the variable len2, but this variable wasn't declared in the example. Sure, it's trivial it should be done but examples that aren't self contained are much less useful in a textbook.

On page 55, 5 lines from the end of the page it says
"operator=() is not assigning a cv::Mat or a cv::Mat (as it might appear),"
When what should be written is
"operator=() is not assigning a cv::Mat TO a cv::Mat (as it might appear),"

On page 47, the text description "Two-dimensional arrays by type with
initializtion value (size in sz)" has a missing letter 'a' in the word initializtion

This is not an exhaustive list of errors I noticed, just some I took note of.
This 2nd edition of the book was very well needed as the previous edition is completely redundant due to the major changes when opencv became more c++ and less c.
It's a shame the preview edition is so riddled with mistakes.
I am hoping this will be fixed by the final edition, as I've paid for the book and would enjoy using it without having to double-check the validity of the content.

Regards,
Yehuda