I'm trying to get a bounding box for the word "ЛИЛИЯ" in this image, using opencv.
(source: litprom.ru)
I am already experimenting with cv::findContours() and different thresholding alogrithms for couple of days, but can not get any satisfying results.
So, what do I know about this word:
letters are of similar size;
letters' height is in range: 40px — 90px;
word is oriented horizontaly (±5˚);
there is one and only one word on this image;
this word does not intersect image's border (it's fully visible);
different parts of image may have different luminosity;
hotspots (totally white areas) may be present on an image.
English is not my native language, so I'm sorry if the question is not properly explained.
If someone needs more images to answer this question, I have at least a dozen more.
Check out stroke width transform. That is used to text detection.
You can preprocess your image with adaptiveThreshold. You should use a blocksize a little bit bigger than your biggest character. I tried on your image with 91 and it gave good results. Then you can use FindContours and filter the blobs/contours using their height. Note that the letters will still be connected one to another so you cannot really filter using the width.
Related
We are developing an app where we need to crop an image according to the selecting object area. User will draw a line and we need to select the object and crop it .This crop need to be like the app: YourMoji
So far we have tried to get the color of the pixels along the line and then comparing those with the color of every pixel in the image and making a path from it to clip the image. But the almost going no where.
Is it possible through this way to crop an image or we are going in the wrong way? Can anyone provide a way to do this Or suggest a way to modify the way we have worked so far? Any advice and suggestions will be greatly appreciated!
Thanks in advance.
I guess what you want is the image segmentation algorithm called Graph Cut.
Here are two Github repositories, hope these would help:
GraphCut
GrabCutIOS
I'm not exactly clued up on image manipulation, but the first algorithm that comes to mind is something like this:
Take the average of the pixels in the line (as you have)
Since you appear to want faces, you might want to weight reds and blues over green. Not much green in faces of any skin tone.
For each pixel, if the colour is within a given threshold outside of your selected average, remove it / make transparent.
Perhaps the closer to the original line (or centroid), the less strict the threshold becomes.
I'd then provide the user with some tools for:
Sensitivity: how large the threshold is
Eraser: to remove parts of the image that your algorithm missed
Paintbrush: to replace parts of the image that your algorithm incorrectly removed.
I've been trying to work on a small hobby project that involves plotting players' positions from a game onto a heatmap, to see where the most active areas are at various points in time.
I'm a bit new to OpenCV and its tools, but I've managed to successfully run some text matching and extraction on the scoreboard and timers in the game, now trying to take the characters' positions from the in-game minimap.
It looks like this, which is the biggest resolution image I'm able to get with (about 185x185):
I'm trying to obtain the positions of only two things: the characters (big circles) and "wards", which are represented by these icons:
So given the assets to them, I thought that because there was too much "noise" in the source image, I'd try to subtract the background of in game minimap from its image, and then try to pattern match the original character and ward image with the resulting image together (which is meant to be the minimap, minus its background). But that didn't even get close to working as you can see:
> >
Even if that did work, I wouldn't be really sure how to handle cases where the icons are partially covering each other, or how I could obtain the positions of those little ward markers.
I'd really appreciate some help, as I've been searching the Internet and banging my head for a few days and haven't gotten anywhere. I've tried a bunch of difference techniques, read guides and articles, and tried a few GUI tools to experiment with but haven't gotten any closer to a method to work this out.
Please help me with what techniques I could or should be using instead, to get the locations of all the characters and wards.
I'm not an OpenCV user, but I can speak to some general problems.
First and foremost, you goofed in subtracting the background map. It appears that you did a straight, arithmetic subtraction of the map's RGB values. For instance, the blue-team icons in the lower-left corner are roughly #99FFFF, and you're subtracting the grayish background of maybe #D0D0FF. This leaves you with #002F00, a very dark green.
Also note that you're subtracting the original map, not the part that shows. Paths beyond view are shaded, but you appear to subtract the original value.
What you need to subtract is a masked background. Unfortunately, building that mask means that you have to find the icons. Masking won't work well at this stage.
Back to the subtraction: don't just blindly subtract. Rather, look for a match in hue. When you find a hue match, simply set that pixel to 0. You have two special cases to watch: icons on the background of their own colour, especially for the blue team. In this case, you need to define the region boundaries.
Start from a pixel that's an exact match to the original background. It won't be shaded, since all such problem pixels are in plain sight of an icon. Expand from that pixel so long as you have the exact match to the original background colour. That will give you the region you can blank out.
Your next problem is to identify icons. You should now have a map with only icons, many of which are fully revealed. Those are easy matches; identify and subtract them, one key icon at a time.
You now have a map of partial icons. Switch the match algorithm: a key icon is now a match to either the exact color, or to black (indicating it was previously covered). Iterate until you have no more matches.
This does still leave you with one problem: an icon that no longer has enough pixels showing to identify. These will be icons that were either entirely covered, or covered except for a small portion that is not unique, such as a few pixels of a red circular border.
For this, a general approach is to keep track of game progress to a small extent: from an earlier time, you know where the icon used to be. Track each icon as a software object. If other icons cover it, assume it's still there until you discover otherwise.
This will handle most cases. You'll still have some problems with minions or sensors that get shot out from underneath a legend's icon, but I trust that your heat map application is not so fragile as to take modelling damage from that situation. The legend will move soon enough, revealing the small item's death. A moving minion isn't covered by a legend for long; they don't move with the same intelligence.
I'm working on a jigsaw puzzle webapp, and one of the requirements is automatically generating puzzle pieces from any image. I'm using RMagick for the image processing. I've got some sets of blank puzzle pieces to use as masks, and I can handle that part, but then I need to trim the whitespace (er, transparentspace) out of the resulting images.
Now, I know I can use trim for this - I might have to put a one-pixel border on it to make sure all four corners are the right color, but that's easy and I can just subtract one pixel from the final number. The only problem is that I also need to record the position of the piece. According to the documentation on trim, the function will "retain the offset information", which sounds like exactly what I need. But I can't find anything about how to retrieve the offset information! Does anyone know how to do that?
If worst comes to worst, I suppose I could always just look through pixel-by-pixel, find the boundaries myself, and use crop to trim the picture, but that wouldn't exactly be good for performance.
Aha, found it. image.page.x and image.page.y give the upper left corner, and then image.rows and image.columns have the height and width.
I am using 4-Way floodfill algorithm.
I have a transparent image with black out line.
That is staring point image(without color).
And after filling the color in this image it look like this
Please help me and let me know what can i do for proper fill.
I used and implemented myself FloodFill in other projects and the algorithm goes trough the whole draw, looking for closed spaces and then draw inside (or outside) them.
Your problem happens with every tool in the world that fills a draw, and the problem is the same, the spaces are not 100% closed.
The floodfill algorithm goes pixel by pixel and when it detect a black pixel, it stops. For example, the arm of the scuba driver is not thick enough or it has holes on it, and the flood fill algorithm manages to go trough it and not detect it as an empty space.
Nobody here can tell you why unless we take your project and analyse it, so the best I can offer is a guideline about where your error could be.
I tried the code with an image that has a very precise defined border around it (from here) and it seems to work OK with that image. I suggest perhaps that if you zoom into your image that there is some grey aliasing around the edges which won't get filled. Perhaps the algorithm has a threshold function that can be tweaked?
Try setting the andTolerance value (I tried 4 which seemed to improve my example).
//Call function to flood fill and get new image with filled color
UIImage *image1 = [self.image floodFillFromPoint:tpoint withColor:newcolor andTolerance:4];
Can anyone share a sample code to draw a non-rectangular part of a picture in delphi canvas?
You're looking for GDI paths. Start here, which explains what paths are in this context, and provides links on the left to explain the functionality available with them.
Google can turn up lots of examples of using paths in Delphi. If you can't find them, post a comment back here and I'll see what I can turn up for you.
Your question is pretty vague. But I suspect what you are looking for is clipping regions. Read up on them. Set the clipping region on the target device to the shape you want, and then draw the image onto the device. Only the part of the image that would be within the clipping region will be drawn.
Canvas.Ellipse(0, 0, 10, 20); // not a rectangle
I use so called runlists for this feature (generalized shapes and blitting them). I've seen them called warplists too. A shape is encoded as a runlist by defining it as a set of horizontal lines, and each line is two integer values (skip n pixels,copy n pixels).
This means you can draw entire lines, leaving you with only "height" draw operations.
So a rectangle is defined (the first "skip" pixels from top level corner to the left corner (xorg,yorg). The rectangle is width_rect wide, and width_pixels goes a line further. width_pixels can be wider than the width of the picture (alignment bytes)
(yorg*width_pixels+xorg , width_rect),
(width_pixels-width_rect , width_rect),
(width_pixels-width_rect , width_rect),
(width_pixels-width_rect , width_rect),
..
..
This way you can make your drawing routines pretty generic, and for simple, regular shapes (rects, circles) it takes only minor math to precalculate these lists. It simplified my shape handling enormously.
However I draw directly to bitmaps, not to canvasses, so I can't help with that part. A primitive that efficiently draws a row, and a way to extract a row from a graphic should be enough.