How to sum pixels from each sliding window in a picture?

Question

Aneta Chwala 2018-12-11

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/435143-how-to-sum-pixels-from-each-sliding-window-in-a-picture

评论： Guillaume 2018-12-12

I have a picture and a sliding window on it. Sliding window is moving and ectracts the number of pixels relevant to setpoint rgb intervals. But I need the whole sum from the picture. How to calculate that ?

3 个评论
显示 1更早的评论隐藏 1更早的评论

Guillaume 2018-12-11

Do the sliding windows overlap?

What is the number of pixels relevant to setpoint rgb intervals?

Aneta Chwala 2018-12-11

my picture size: 3133x2375x3 uint8

sliding window size: 64x64 with step 64

The sliding picture moves starting with left up corner and ending at right down corner.

I have two variables A and B which are numbers of pixels for rgb intervals for A it's: (R > 45 & R < 180), (G > 50 & G < 185), (B > 160 & B < 215), and for B: (R > 40 & R < 115), (G > 6 & G < 80), (B > 10 & B < 75).

I need a sum of A and B from every slide (which is from the whole picture).

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Guillaume 2018-12-11

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/435143-how-to-sum-pixels-from-each-sliding-window-in-a-picture#answer_351737

在 MATLAB Online 中打开

There are many ways you can apply the sliding window. For contiguous, non-overlapping blocks you can use blockproc. If the blocks overlap you can use nlfilter or colfilt, or even simpler since your operation is simply a sum conv2. In any, case before any of these, you can first calculate your A and B per pixel:

rgbAlower = uint8(cat(3, 45, 50, 160));    %lower thresholds stored across pages
rgbAupper = uint8(cat(3, 180,  185, 215)); %upper thresholds stored across pages
A = all(yourimage > rgbAlower & yourimage < rgbAupper, 3);  %is pixel within threshold. Requires R2016b or later

Applying blockproc to that is trivial

result = blockproc(A, [64 64], @(bloc) sum(bloc.data(:)))  %with suitable options for what you want to do at the edges

or with a convolution

fullconv = conv2(A, ones(64, 64), 'valid');  %window of 64x64 with step of 1
result = fullconv(1:64:end, 1:64:end)        %extract blocks with step of 64