4.9 | 32 ratings Rate this file 187 downloads (last 30 days) File Size: 269.05 KB File ID: #4551

inpaint_nans

29 Feb 2004 (Updated 28 Jun 2006)

Interpolates (& extrapolates) NaN elements in a 2d array.

Editor's Notes:

This is a File Exchange Select file.

Select files are submissions that have been peer-reviewed and approved as meeting a high standard of utility and quality.

Download Now | Watch this File

File Information
Description	Interpolate NaN elements in a 2-d array using non-NaN elements. Can also extrapolate, as it does not use a triangulation of the data. Inpaint_nans offers several different approaches to the interpolation, which give tradeoffs in accuracy versus speed and memory required. All the methods currently found in inpaint_nans are based on sparse linear algebra and PDE discretizations. In essence, a PDE is solved to be consistent with the information supplied.
Acknowledgements	This submission has inspired the following: Inpainting nan elements in 3-d
MATLAB release	MATLAB 5.2 (R10)
Other requirements	Written in version 5.2, tested up to release 14 SP1
Zip File Content
Published M Files	inpaint_nans_demo
Other Files	Inpaint_nans/.DS_Store, Inpaint_nans/garden50.jpg, Inpaint_nans/inpaint_nans.m, Inpaint_nans/monet_adresse.jpg, Inpaint_nans/demo/.DS_Store, Inpaint_nans/demo/inpaint_nans_demo.m, Inpaint_nans/doc/Nomination comments.rtf, Inpaint_nans/doc/methods_of_inpaint_nans.m, Inpaint_nans/test/.DS_Store, Inpaint_nans/demo/html/inpaint_nans_demo_05.png, Inpaint_nans/demo/html/inpaint_nans_demo_06.png, Inpaint_nans/demo/html/inpaint_nans_demo.png, Inpaint_nans/demo/html/inpaint_nans_demo_01.png, Inpaint_nans/demo/html/inpaint_nans_demo_02.png, Inpaint_nans/demo/html/inpaint_nans_demo_03.png, Inpaint_nans/demo/html/inpaint_nans_demo_04.png, Inpaint_nans/test/test_main.m

Tags for This File
Everyone's Tags	approximation(3), extrapolates(3), file exchange select, inpaint, interpolation, nan, triangulation
Tags I've Applied
Add New Tags	Please login to tag files.

Comments and Ratings (38)
03 Mar 2004	Urs Schwarz (us)	A very useful and meticulously coded, data-driven function. A must to look at and learn from for anybody who is working with incomplete 2D data sets, in particular, dirty images.
28 Jun 2004	Evan Zamir	Nice! Especially method 4 using the virtual springs.
10 Aug 2004	Tom Reiser	I love it!
29 Oct 2004	d d
30 Oct 2004	jing tian	excellent!
06 Dec 2004	gok gok	how to handle non-rectangular areas? Triangle for example. I'm getting holes from DELAUNAY and your routine will be very helpful to fix this bug.
14 Feb 2005	K. Sturm	Well done !
16 Jun 2005	Cris C.	Really a good work. Thank you.
19 Apr 2006	sacrosancttayyar sacrosancttayyar
21 Apr 2006	Rajeev Srivastava	excellent work.
21 Apr 2006	Ian Howat	very fast, thanks!
31 Aug 2006	Mike Kadour	Just what I was looking for! Many thanks!
13 Oct 2006	mario fiorini	great!
17 Jan 2007	Amar chawla	Great piece of work! really helped....
24 Sep 2007	Emily Spahn	Thank you! This does exactly what I was hoping it would!
31 Oct 2007	martin klöppel	realy useful
30 Dec 2007	Sara Alexander	Really helpful with minimal fuss!
08 Feb 2008	D. G.	Simply excellent. Nothing to declare.
08 Mar 2008	Carlos Adrián Vargas Aguilera	Here you have my 5 stars... Congratulations, and thanks for sharing it...
16 Apr 2008	Bob Field	Magical
13 Aug 2008	E E	Comment=[T H A N NaN Y O U V E NaN Y M NaN NaN H]; [M,N] = size(Comment); for k = 1:length(M); Comment=inpaint_nans(Comment); end Comment= T H A N K Y O U V E R Y M U C H
16 Sep 2008	Sherif Abd El-Gawad	Thanks a lot and Good Contribution
07 Oct 2008	Frederic Moisy	Thank you once again John, exactly what I needed. Please TMW: include this in next release of matlab!
16 Dec 2008	M P
16 Dec 2008	Kenneth Eaton
19 Dec 2008	M P	The big question is: what is the meaning of del^2? I've other questions. Method zero. First example: my input matrix is A=NaN A = NaN nn = Empty matrix: 0-by-2 ??? Improper assignment with rectangular empty matrix. A matrix 1x1, only one element, is not a 2D matrix in your interpretation? The error is in "sub2ind" command; it has "nn" in input. Second example: my input matrix is A=9 A = 9 neighbors_list = [] neighbors_list = [] all_list = [] ??? Index exceeds matrix dimensions. The error is in the "find" command. 3° example: my input matrix is a 2x2 A = NaN 1 NaN 2 B = NaN NaN 1 2 Warning: Rank deficient, rank = 0. ??? Improper assignment with rectangular empty matrix. The error is in "fda" command. 4° example: my input matrix is 3x2 A = NaN 1 NaN 2 NaN 3 B = NaN NaN NaN 1 2 3 Warning: Rank deficient, rank = 1. In this case not error but warning and the row is the same of the 3° case ("fda" command). 5° example: input matrix is 3x3 A = NaN 1 1 NaN 2 2 NaN 3 3 B = 1.0000 2.0000 3.0000 1.0000 2.0000 3.0000 1.0000 2.0000 3.0000 In this case the output B is ok; my question is: why a triple NaN is just 1 2 3? 6° example: input matrix is 3x3, I'm finding to understand the strategy of NaNs substitution. A = NaN 1 4 NaN 2 5 NaN 3 6 B = -2 -1 0 1 2 3 4 5 6 Ok, program is ok. Not error. Why NaNs value are -2 -1 0? Other important question: why my input A (3x3) is now an output B (9x1)? 7° example: input matrix 10x10 of zeros with only one NaN in (5,5) position. A = 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 NaN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 B = 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Output matrix is a vector 100x1 with all zeros. Why NaN is just zero? 8° example: like the 7° but with A of ones A = 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 NaN 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 B = 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I've obtained NaN=1. So I try with an image, because in the image's file there are 3 images. 9° example: matrix 3x3x3 A(:,:,1) = NaN 1 4 NaN 2 5 NaN 3 6 A(:,:,2) = NaN 1 4 NaN 2 5 NaN 3 6 A(:,:,3) = NaN 1 4 NaN 2 5 NaN 3 6 B = -2.0000 <-- why negative? -1.0000 <-- why negative? 0 <-- why? 1.0000 2.0000 3.0000 4.0000 5.0000 6.0000 2.5000 <-- why .5? 3.5000 <-- why .5? 4.5000 <-- why .5? 1.0000 2.0000 3.0000 4.0000 5.0000 6.0000 2.5000 <-- why? 3.5000 <-- why? 4.5000 <-- why? 1.0000 2.0000 3.0000 4.0000 5.0000 6.0000 In your vision an image is 2D? I think is affirmative answer because program is in 2D but if is so the triple color-profundity is a 3=1 formula? How is possible to show an image if NaN is transformed in a number with decimal numbers? NaN value is or is not a number? If yes, then I can put in the NaN position any other number? NaN is 0/0, is certainly a number but PC don't say what number is it. Then is all numbers, in my vision. Is possible it is 1 but also 2 and 3 etc etc. ... NaN=-1 --> N=2 NaN=0 --> N=1 --------> this is half X NaN=1 --> N=0 --------> this is the other half X NaN=2 --> N=-1 NaN=3 --> N=-2 ... NaN=10 --> N=-9 ... N=Number is my input. And NaN is a natural bent of PC. Say to me, if you want. Regards CU
19 Dec 2008	V. Poor
19 Dec 2008	alain boyer
19 Dec 2008	Kenneth Eaton	@Marco, You appear to have completely missed the point of this function. It is designed to REMOVE NaNs appearing in a matrix, overwriting those values with interpolated values obtained from neighboring entries. You seem to think that John is arbitrarily redefining what NaN is... he is not, because that would be a silly waste of time. The numbers he computes to REPLACE the NaNs are approximated from the surrounding values with a number of different methods. As far as some of the errors you got, John could probably place error checks in the function to allow for a more graceful and informative exit. However, there's no reason you would/should ever enter some of those arguments in the first place. For example, entering a scalar NaN is silly, since the program has no nearby non-NaN elements to use for extrapolating replacement values. This submission certainly doesn't deserve the low rating you gave it, especially since the misunderstanding lies with your own misinterpretation of what it does.
19 Dec 2008	M P	"is arbitrarily redefining what NaN is"... no, I don't think this. I would understand what is "neighboring" strategy; if I've an image with four azure points and I put a NaN in one of this points the program answer is the transformation of NaN in an azure point or a near color azure? The NaN point with "image" command is a black point and an image with 3 azure points and a black point is an original image so. If I change the NaN point with a number I can choose a value in the Inf sector of the values. This I say. And I repeat the question: what is neighboring strategy? Is del^2. Can I have more informations? "surrounding values"..... surround=margin..... can you write one of my examples with explanations? "different methods".... I'm speacking of zero method, the default method "entering a scalar NaN is silly".... why? Or also, what number is not-silly? Take an example. "misunderstanding lies with your own misinterpretation of what it does".... lies? where are? Interpretation..... the comment is genuine
19 Dec 2008	Kenneth Eaton	@Marco, If you want more specific information about the methods than what is in the help documentation, you would have to either delve into the code itself or have John explain it to you. Also, the word "lies" is a conjugation of the verb "to lay", not a noun meaning "falsehoods". To rewrite my statement in another way: "...the misunderstanding is a result of your own misinterpretation."
14 Jan 2009	C Schwalm	John, I saw in your zip file how you are thinking of removing method 5. Let me make a counter suggestion: expand it. I can think many of reasons and applications where a neighborhood average would be useful. In terms of expansion: Well, you could allow for the size of the neighborhood, its orientation (forward, backward, center), and the function (sum, mean, var, mode, etc.) to be passed as additional arguments. Either way, this function is very useful and I use it often in place of standard Matlab functions.
24 Mar 2009	Karel
28 Mar 2009	Stuart
25 May 2009	Stefan
25 May 2009	Stefan	Hello, I recently started studying about inpainting and i ran across Mr. D'Errico's program and I have to say it's pretty cool, tried it on a few pics of mine and i wase pleased with the results. I need some help about understanding the basic theory on the subject, how the elliptic PDE's are implemented in inpainting etc. and if anyone could give me a hint on where i could find some books, documents on the matter i would be very grateful. Regards
09 Jun 2009	Andrey Rubshtein
28 Sep 2009	Jon

Please login to add a comment or rating.

Updates
18 Jul 2005	This version of inpaint_nans will also handle inpainting (interpolating/extrapolating) simple vectors with NaN elements.
17 Apr 2006	Release 2: Bug fix for method 5. Clean up help. Add HTML, screenshot, examples, demo comparison to griddata.
28 Jun 2006	Created Select directory structure

Tag Activity for this File
Tag	Applied By	Date/Time
approximation	John D'Errico	22 Oct 2008 07:15:13
interpolation	John D'Errico	22 Oct 2008 07:15:13
extrapolates	John D'Errico	22 Oct 2008 07:15:13
nan	John D'Errico	22 Oct 2008 07:15:13
triangulation	John D'Errico	22 Oct 2008 07:15:13
approximation	Patrick HannaSecure	31 Oct 2008 14:25:15
extrapolates	Patrick HannaSecure	31 Oct 2008 14:25:21
inpaint	Oliver Woodford	26 Jan 2009 19:02:25
file exchange select	Oliver Woodford	27 Jan 2009 08:05:11
extrapolates	Seth	18 Jun 2009 14:58:49
approximation	Seth	18 Jun 2009 14:58:52

MATLAB Central Terms of Use

NOTICE: Any content you submit to MATLAB Central, including personal information, is not subject to the protections which may be afforded information collected under other sections of The MathWorks, Inc. Web site. You are entirely responsible for all content that you upload, post, e-mail, transmit or otherwise make available via MATLAB Central. The MathWorks does not control the content posted by visitors to MATLAB Central and, does not guarantee the accuracy, integrity, or quality of such content. Under no circumstances will The MathWorks be liable in any way for any content not authored by The MathWorks, or any loss or damage of any kind incurred as a result of the use of any content posted, e-mailed, transmitted or otherwise made available via MATLAB Central. Read the complete Terms prior to use.