{"id":974,"date":"2014-02-04T13:28:43","date_gmt":"2014-02-04T18:28:43","guid":{"rendered":"https:\/\/blogs.mathworks.com\/steve\/?p=974"},"modified":"2019-11-01T10:59:24","modified_gmt":"2019-11-01T14:59:24","slug":"regionprops-tables-and-struct2table","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/steve\/2014\/02\/04\/regionprops-tables-and-struct2table\/","title":{"rendered":"regionprops, tables, and struct2table"},"content":{"rendered":"

Today I was talking to Spandan (a regular guest blogger here) about the history of the Image Processing Toolbox function regionprops<\/tt>. This function is a widely-used work-horse of the toolbox. Many of my blog posts here have shown it in practice.<\/p>

The function regionprops<\/tt> computes properties of regions of pixels. Most often, these regions correspond to individual \"objects\" (connected components of foreground pixels) of a binary image. Here's an example.<\/p>

bw = imread('text.png'<\/span>);\r\nimshow(bw)\r\n<\/pre>\"\" 
s = regionprops(bw)\r\n<\/pre>
\r\ns = \r\n\r\n88x1 struct array with fields:\r\n\r\n    Area\r\n    Centroid\r\n    BoundingBox\r\n\r\n<\/pre>
The output is an 88x1 struct array, elements of which contain measurements for objects in the binary image. (We didn't tell regionprops<\/tt> what measurements to compute, so it computed Area, Centroid, and BoundingBox by default.) Here are the measurements for the 10th object.<\/p>
s(10)\r\n<\/pre>
\r\nans = \r\n\r\n           Area: 61\r\n       Centroid: [54 39]\r\n    BoundingBox: [49.5000 33.5000 9 11]\r\n\r\n<\/pre>
This form of output is convenient in some ways but inconvenient or awkward in others. For example, it requires the obscure \"comma-separated-list\" syntax to get all the area measurements into a single vector to call a function such as mean<\/tt>, min<\/tt>, or max<\/tt>.<\/p>
a = [s.Area];\r\nmean(a)\r\n<\/pre>
\r\nans =\r\n\r\n   54.2386\r\n\r\n<\/pre>
min(a)\r\n<\/pre>
\r\nans =\r\n\r\n     6\r\n\r\n<\/pre>
max(a)\r\n<\/pre>
\r\nans =\r\n\r\n   106\r\n\r\n<\/pre>
In the comma-separated list syntax for struct arrays, the expression [s.Area]<\/tt> automatically expands into the expression [s(1).Area, s(2).Area, ..., s(end).Area]<\/tt>. But relatively few people recognize or understand this syntax.<\/p>
Another inconvenient aspect of the struct array aspect is that it's hard to get a tabular view of all the data. As shown above, you can see the data for the 10th object by displaying s(10)<\/tt>, but it's hard to see all the data for multiple objects. For example:<\/p>
s(1:5)\r\n<\/pre>
\r\nans = \r\n\r\n5x1 struct array with fields:\r\n\r\n    Area\r\n    Centroid\r\n    BoundingBox\r\n\r\n<\/pre>
If I were designing regionprops<\/tt> from scratch today, based on the most recent version of MATLAB, I would have regionprops<\/tt> return a table<\/tt>. Tables were introduced as a new fundamental MATLAB type in R2013b. Here's a simple example to illustrate.<\/p>
age = [38 43 38 40 49]';\r\nheight = [71 69 64 67 64]';\r\nweight = [176 163 131 133 119]';\r\nblood_pressure = [124 93; 109 77; 125 83; 117 75; 122 80];\r\nt = table(age,height,weight,blood_pressure)\r\n<\/pre>
\r\nt = \r\n\r\n    age    height    weight    blood_pressure \r\n    ___    ______    ______    _______________\r\n\r\n    38     71        176       124          93\r\n    43     69        163       109          77\r\n    38     64        131       125          83\r\n    40     67        133       117          75\r\n    49     64        119       122          80\r\n\r\n<\/pre>
t.height<\/tt> is an ordinary 5-by-1 vector of heights.<\/p>
t.height\r\n<\/pre>
\r\nans =\r\n\r\n    71\r\n    69\r\n    64\r\n    67\r\n    64\r\n\r\n<\/pre>
t.blood_pressure<\/tt> is an ordinary 5-by-2 matrix. (Table \"variables\" are often column vectors, but they can have any size as long as the number of rows is the same as the height of the table.)<\/p>
t.blood_pressure\r\n<\/pre>
\r\nans =\r\n\r\n   124    93\r\n   109    77\r\n   125    83\r\n   117    75\r\n   122    80\r\n\r\n<\/pre>
We are excited about the introduction of tables in MATLAB, and we think they will gradually be used in more and more situations where previously people used struct arrays.<\/p>
Such as regionprops<\/tt>.<\/p>
So what to do?<\/p>
Well, the function struct2table<\/tt> comes in very handy here! As you might have guessed from the name, it automatically converts a struct array into a table.<\/p>
Try it:<\/p>
t = struct2table(s)\r\n<\/pre>
\r\nt = \r\n\r\n    Area         Centroid         BoundingBox \r\n    ____    __________________    ____________\r\n\r\n     66         11        13.5    [1x4 double]\r\n     41     7.6829      38.171    [1x4 double]\r\n     63     17.889      38.857    [1x4 double]\r\n     80      22.95      15.963    [1x4 double]\r\n     53     26.472      36.604    [1x4 double]\r\n     63     34.889      16.857    [1x4 double]\r\n     63     34.889      38.857    [1x4 double]\r\n     41     43.683      38.171    [1x4 double]\r\n     48     48.396      15.875    [1x4 double]\r\n     61         54          39    [1x4 double]\r\n     63     56.889      16.857    [1x4 double]\r\n     41     65.683      16.171    [1x4 double]\r\n     48     67.396      37.875    [1x4 double]\r\n    105      78.79       16.79    [1x4 double]\r\n     66       76.5          39    [1x4 double]\r\n     68     93.529      39.265    [1x4 double]\r\n     92        100      16.696    [1x4 double]\r\n     41     106.68      38.171    [1x4 double]\r\n     68     113.53      17.265    [1x4 double]\r\n      6        114        31.5    [1x4 double]\r\n     33        114          39    [1x4 double]\r\n     85     123.24      38.259    [1x4 double]\r\n     48      121.4      15.875    [1x4 double]\r\n     63     129.89      16.857    [1x4 double]\r\n    100     134.87       40.86    [1x4 double]\r\n     41     138.68      16.171    [1x4 double]\r\n     63     145.89      38.857    [1x4 double]\r\n     61        149          17    [1x4 double]\r\n     80     159.95      15.963    [1x4 double]\r\n     85     163.26      185.76    [1x4 double]\r\n      6      156.5         195    [1x4 double]\r\n      6      156.5         211    [1x4 double]\r\n     85     163.26      218.76    [1x4 double]\r\n     48      159.4      37.875    [1x4 double]\r\n     61        164         107    [1x4 double]\r\n     68     164.26      117.47    [1x4 double]\r\n     63     163.86      129.11    [1x4 double]\r\n     41     163.17      139.32    [1x4 double]\r\n     68     164.26      147.47    [1x4 double]\r\n     61        164         164    [1x4 double]\r\n     63     163.86      175.11    [1x4 double]\r\n     33        164         195    [1x4 double]\r\n     54     163.44         203    [1x4 double]\r\n     33        164         211    [1x4 double]\r\n     80     167.95      37.962    [1x4 double]\r\n      9        168         233    [1x4 double]\r\n      9        168         238    [1x4 double]\r\n      9        168         243    [1x4 double]\r\n     63     171.89      16.857    [1x4 double]\r\n     68     180.53      39.265    [1x4 double]\r\n     48     184.88      67.604    [1x4 double]\r\n    106      183.6      118.53    [1x4 double]\r\n      6      178.5         127    [1x4 double]\r\n     85     185.26      134.76    [1x4 double]\r\n     85     185.26      164.24    [1x4 double]\r\n     85     185.26      179.76    [1x4 double]\r\n      6      178.5         212    [1x4 double]\r\n     85     185.26      238.76    [1x4 double]\r\n     85     184.24      16.259    [1x4 double]\r\n     72     185.68      76.653    [1x4 double]\r\n     63     185.86      88.111    [1x4 double]\r\n     41     185.17      98.317    [1x4 double]\r\n     63     185.86      107.11    [1x4 double]\r\n     33        186         127    [1x4 double]\r\n     74      187.2      152.31    [1x4 double]\r\n     63     185.86      192.11    [1x4 double]\r\n     72     185.68      203.65    [1x4 double]\r\n     33        186         212    [1x4 double]\r\n     68     186.26      219.47    [1x4 double]\r\n     41     185.17      230.32    [1x4 double]\r\n     48      188.4      37.875    [1x4 double]\r\n      9        199          43    [1x4 double]\r\n     48     206.88       161.6    [1x4 double]\r\n      6      200.5         235    [1x4 double]\r\n      9        204          43    [1x4 double]\r\n     61        208         125    [1x4 double]\r\n    105     207.79      139.21    [1x4 double]\r\n     63     207.86      153.11    [1x4 double]\r\n     61        208         170    [1x4 double]\r\n     74      209.2      181.31    [1x4 double]\r\n     61        208         192    [1x4 double]\r\n     41     207.17      207.32    [1x4 double]\r\n     63     207.86      216.11    [1x4 double]\r\n     54     207.44         227    [1x4 double]\r\n     33        208         235    [1x4 double]\r\n     41     207.17      242.32    [1x4 double]\r\n      9        209          43    [1x4 double]\r\n      9        212         117    [1x4 double]\r\n\r\n<\/pre>
Immediately, you can see that the data is much more easily visible with a table. You get scrolling and editing if you open t<\/tt> in the Variable Editor.<\/p>
\"\" <\/p>
You can plot directly with the data in a table variable:<\/p>
imshow(bw)\r\nhold on<\/span>\r\nplot(t.Centroid(:,1),t.Centroid(:,2),'r*'<\/span>)\r\naxis([.5 100.5 .5 100.5])\r\nhold off<\/span>\r\n<\/pre>\"\" 
Logical indexing comes in handy for filtering objects based on various criteria. For example, the following code removes objects with area less than 10 from the table.<\/p>
t(t.Area < 10,:) = [];\r\n<\/pre>
And you can easily export tables to various formats. Here is our table exported to a .CSV file and then read into Excel.<\/p>
writetable(t,'regionprops.csv'<\/span>);\r\n<\/pre>
\"\" <\/p>
I encourage everyone who has used regionprops<\/tt> to play around with tables (assuming you have MATLAB R2013b) to see how they might be useful in your own work.<\/p>