why reshape command is actually needed ?

why reshape command is actually needed in image segmentation

[John D'Errico]

Impossible to answer. Reshape is useful and even necessary in a huge number of places. In your case, it was apparently needed because an array was the wrong shape for the next procedure.

But if you really want a better answer, you need to tell people what you are doing, and far more clearly. ACTUALLY show the operation. Then tell us what shape the input was before the reshape.

https://www.mathworks.com/matlabcentral/answers/6200-tutorial-how-to-ask-a-question-on-answers-and-get-a-fast-answer

[Touhidul islam]

In an image processing code ,one coder has converted 218 180 3 images to 39240 3 by using reshape . why he did this he hasn't explained.

[Walter Roberson]

reshape() to something by 3 would be done as preparation for clustering by color, since the clustering routines expect the variables to go across the columns and that each row is a different sample.

[Touhidul islam]

sorry , i couldnot understand what you have tried to say. Please, can you again explain?

Answer 1

[Walter Roberson]

reshape() is never mandatory. There is always a way to write for loops to copy the data into the desired new shape. reshape() is just fast and convenient.

Answer 2

[Star Strider]

@Touhidul islam —

With the later advantage of being able to see what are apparently the entire code, what you describe is the ‘F’ matrix, formed by reshaping an image. The purpose (from the code you posted earlier today in Can anyone help me out with the following code ? (link)) appears to be to provide an easy reference for replacing NaN values by a specific random row of ‘F’ later in the code.

It would be of significant benefit to us if you would post all the relevant code so we can see it and provide you with the information you want.