I am writing a paper for a scientific journal and I want to include images that I made with Matlab. The way I usually do this is by plotting the image in Matlab and interactively go to the plotting window in 'File->Save as..' and save the image as an eps. When I include the image in my paper (which I write with LaTeX) the image looks correct overall (there are no artifacts) but it does not look as sharp as on my screen. Is there a better way to save the images with better quality so that they do not look blurry when I insert them in LaTeX?
Thank you.
Related
Example Image
For example as Image above, area I need is just like red box, and other section doesn't have any labels for classification/object detection.
What I think is "If I use cropped image to red box will occur better effect" because there's to much useless area without labels in original image. When mosaic augmentation used in yolo v4, It will put images together in one. And, because there's so many area without labels, data after mosaic can be useless than before.
But, This is just my guess, and I need a test to confirm it, but the lack of computing power is limiting the actual test. So the question is, Is it possible to actually improve performance if the original image is cropped in the form of a red box? Is that why I guessed correctly?
Also, my partner said that cropping is not a good choice in Yolo because it can ruin the proportion of the object, but I couldn't understand what the proportion of the object meant in Yolo. I wonder why the proportion of objects in Yolo is not suit with cropping.
Thanks for read, and have a nice day
simply you shouldn't resize the images, however, if the training/testing data set contains considerable difference among width and heights, use the data augmentation methods. Pls, follow the link for more information.
https://github.com/pjreddie/darknet/issues/800
been trying to find the answer to why everybody converts an image to grayscale before processing?
For example, this website with instructions teaching people how to build a simple scanning program converts photo to greyscale first before passing commands to manipulate the image itself.
In the second example, this thread on stackoverflow shows a person also converts the image to grayscale before extracting text from his image.
Does this process make the image easier to manipulate? Or does it give better results when extracting text? If so, shouldn't a binary image give the best result in the case of extracting text?
More often than not, grayscale has all the relevant information to complete a particular task. So reducing the image to grayscale greatly simplifies calculations and removes redundancies.
Binary image is great too but it sacrifices too many information for it to be useful in many cases. And most library supports a minimum of 8 bit image processing anyway for a true binary data structure to be useful.
Imagine having to create a program to recognize text on paper. Having a color image doesn't help you to better read the text. The text can be in various color but you can read the text even if its in black and white. You can argue that binary image should also give the same performance and that is true IF there are no noise such as shadow on the paper.
Once there are noise elements exist on the image, you will need more information to separate text from noise and that is when grayscale is useful.
Moreover the most used and reliable information for advanced image processing is the edges and its textures. Both which can be obtained from a grayscale image.
What is the best way to store images for web-sites?
What size I should not exceed?
Now, I save all interface-files in png (mostly in sprites) and usual images in jpg's. Some images are about 100-150Kb.
What quality and compression should I choose while saving images?
Preparing Images
To prepare images for presentation on
the Web, follow these steps:
Begin a new file. If it’s a photo or original artwork, use 300 dpi as
the resolution. If you are creating
something simple, set the resolution
to 72 dpi.
Use the drawing, text, shape and other tools provided in the graphics
program to create your image.
Save your file in the native file format (a file type meant for editing
that is specific to the image editing
program) of the program you are using.
For example, a Photoshop document
creates files with a ".psd" extension.
If you need to make a change or edit
later, do it in this file rather than
on the Web.
If necessary, lower the resolution of the image to 72 dpi.
If necessary, resize the image to the height and width in pixels
specifications desired.
Save the file in a Web-ready file format, usually a GIF (most programs
give you several choices for file
types), finding a balance between the
way the file looks and an appropriate
file size. Some image programs will
let you preview the final file output,
whie in others, you have to save it
and load it in the browser to see the
final effect.
Preparing Photographs
To prepare photographs for
presentation on the web, follow these
steps:
Download the photo from your digital
camera or scan in the photograph. The
process for this varies depending on
the software bundled with your scanner
or camera. The image will probably end
up in "Your Photos" on Windows, or the
"Photos" folder on a Mac, unless you
chose some other folder as the default
destination.
Use the color correction, drawing, text, shape, and other tools provided
in the graphics program to alter and
improve your image.
Save a copy of your file in the native file format of the program you
are using. Make later changes or edits
in this file rather than on the web.
With photographs, save an untouched
version of the photo in case you want
to use it again for another purpose.
If necessary, lower the resolution of the image to 72dpi.
If necessary, resize the image to the height and width in pixels
specifications desired.
Save the file in a web-ready format, usually a JPEG, finding a
balance between the way the file looks
and an appropriate file size.
Making images web-ready is all about
compromise. In general, the better
your image looks, the larger the file
size will be. The larger the file
size, the longer it will take to
download. Your job is to find a
compromise between quality and file
size. Some things to consider are:
Is this image conveying crucial information? If so, go for higher
quality.
Is the image on a page with a lot of other images? If so, make the file
size smaller.
Will your web visitor be able to detect that the quality of a photo or
image isn’t very good?
Make images of people fairly high quality because visual acuity is
strongest with people’s faces.
(Source)
For your images, here is a tip:
Do not scale images in HTML
Web page designers sometimes set image
dimensions by using the width and
height attributes of the HTML image
element. Avoid doing this since it can
result in images being larger than
needed. For example, if your page
requires image myimg.jpg which has
dimensions 240x720 but displays it
with dimensions 120x360 using the
width and height attributes, then the
browser will download an image that is
larger than necessary.
(Source, see another tip)
This all depends on what the image is for. If you're generally adhering to a web-safe palette, gifs will come out smaller and clearer than a jpeg. Jpeg's are for photos. For jpegs to handle websafe colors well, you need to save at 88%+ quality, which increases file size. Png's generally have the best quality, and in some cases (when you don't need transparency) can come out smaller than jpgs, and comparable to gif. So, for jpeg photos, 80 quality, (sometimes even 75 for smaller images) will get you decent results and small filesize. Use gifs / png for site imagery generally. As far as images on a page, I try not to exceed 30kb total, assuming the page isn't something that is focused on photos / imagery. Hope this helps.
It depends on what your image is of, and what the quality needs to be, whether you choose quality over page loading etc.
I personally save all mine in .png but IE has a problem rendering the colour correctly in .png so if you are using them for aestetical purposes I would save .jpg
As to the size of the image, like said it depends really on the quality you want your image and how it affects performance if it is a large file.
i usually use save for web mode that exists in photoshop. obviously the lower size is best. The sizes that you talking about (100-150) are good or not in relation to the dimension of the image. Anyway i advice you to use sprites and when possible in jpeg format.
something like 75-80% quality are good enought to a good quality of image
And please pay attention to the cmyk mode, that cause problems in some browsers
I'm trying to get an image of a blackboard readable by OCR. Naturally, most OCR software doesn't like dirty images. What image processing should I try to put the image through to clean the image up?
Have you tried the OCR software yet? It's likely that the OCR software is well suited to reading what's essentially already a black and white image.
However, if you were required to do so you could try to:
Threshold the image.
Essentially take a greyscale version of the image and turn it into black / white pixels
Perform Binary Dilation to grow the remaining objects
Perform Binary Erosion
The idea is by dilating then eroding you would remove any rough / noisy edges and then you can pass the skeletonized image to the OCR.
There are probably plenty of methods to achieve a similar result. Given that there are entire books devoted to computer vision this answer will hardly do them justice.
The only texts I have are from 1997, but surely there's been more written on the subject since.
Algorithms for Image Processing and Computer Vision - J.R. Parker
Digital Image Processing - Gonzalez / Woods
Offhand, I'd say invert the image (reverse the colors, so that the writing is black on white) and increase the contrast a bit. You can try modifying the brightness to get the erased chalk fogginess to disappear into the background.
In Photoshop, the Levels dialog may be your most useful image adjustment. Mimicking this in code is another subject, entirely.
The basis of Levels is that you adjust the max, min and midpoints of the brightness levels. Usually shown on a histogram, you adjust the points such that you obtain the desired amount of contrast, but also move the midpoint such that text in the image is the most well-defined; critical for OCR applications. By moving the midpoint you can "eliminate" the grayscale fuzz that ordinarily surrounds handwriting by causing it to disappear into the light (or dark) areas of the image.
Also you might try converting the image to 1-bit after such an adjustment, forcing everything to black or white. Sometimes this speeds up the OCR process. But be careful, it also will discard detail.
Have you tried edge detection techniques such as Roberts Cross and Sobel operator to filter noise out of the image? Without seeing the quality of the image, can't say how effective that'd be.
Not sure how constrained you are in the choice of OCR solution, but the ABBYY OCR engine (and a web API based on it, http://www.wisetrend.com/wisetrend_ocr_cloud.shtml ) includes automatic image cleanup / texture removal options.
There are commercial solutions but cleaning up board images appears to be an open problem. Add OCR to an unsolved problem, and you get... an unsolved problem.
What are some of the points that I need to follow if I want to have good quality images in a LaTeX document. These images are mostly screenshots of a software application or flow charts.
Below are two such images.
Flow Chart
Screenshot
Thanx
For diagrams, the rule is to use vector formats as much as you can — PDF, EPS or native LaTeX packages. When using vector graphics, the picture does not loose resolution and can be scaled freely. For a flow chart, I would either export it from the drawing application as a PDF, or use PGF/Tikz to produce it from LaTeX (see also examples). If your drawing application does not have a PDF export, consider using one that does — e.g., UMLet.
If you can't use vector graphics (e.g., because it is a screenshot), make sure you use high-enough resolution to begin with. If it is an academic paper, the publisher usually has guidelines for this.
If you use PDFLatex you can use png images and in those cases you definately should use png over jpeg. PNG compression is not lossy, so you get the best quality at the expense of file size.
The second important point is to create the images with sufficient resolution, for printing it should be about 300-600 dpi, higher is better but the filesize of the images and the resulting document will increase. For documents that will only be looked at a screen you can use a lower resolution, about 72-100 dpi should be enough.
For diagrams you should create vector graphics (eps or pdf) if possible, that way you do not lose any quality.
For screenshots, there is not much to do, but for flow charts, I'd suggest to create them in PDF format (vectorized) and to compile your LaTeX source with pdflatex.
for the flowchart i'd suggest TikZ, then your chart is directly typeset in TeX. Here's an example: http://www.texample.net/tikz/examples/simple-flow-chart/
Screenshots are pretty much a lost cause. I've had a good experience saving them as PDF and then embedding them, but you want to make sure you're on a high-res capture to begin with.
Charts are very easy. Most graphics programs (e.g., Vizio, OmniGraffle) will let you save it as EPS or PDF, and scaling works fairly well.