I'm doing a photo app and sometimes the lighting is off in certain areas and the picture isn't clear. I was wondering if there was a feature that can auto adjust the brightness, contrast, exposure, saturation of a picture like in photoshop.
I don't want to manually adjust images like the sample code given by apple:
https://developer.apple.com/library/ios/samplecode/GLImageProcessing/Introduction/Intro.html
I want something that will auto adjust or correct the photo
As an alternative you could use AVFoundation to make your implementation of the camera and set the ImageQuality to high and the autofocus or tap to focus feature. Otherwise, I am almost certain you cannot set this properties, The UIImagePicker controller included in the SDK is really expensive memory wise and gives you an image instead of raw data (another benefit of using AVFoundation). This is a good tutorial for this in case you would like to check it out:
http://www.musicalgeometry.com/?p=1297
Apparently someone has created it on Github: https://github.com/proth/UIImage-PRAutoAdjust
Once imported, I used it the following:
self.imageView.image = [self.imageView.image autoAdjustImage];
Related
I am researching into the best way to detect test in a photo using open source libraries.
I think the standard way is as follows (note: steps 1 - 4 all use OpenCV):
1) detect outline of document
2) transform document so it's flat and cropped, using said outline
3) Make the background of document white, using a filter
4) Feed resulting image to Tesseract
Is this the optimum process, or is there a better way, or better tools?
Also, what happens for case if the photo doesn't have a document outline (It's possible that step 1 & 2 are redundant)?
Is there anyway to automatically detect document orientation (i.e. portrait / landscape)?
I think your process is fine. I've used a similar process for an Android project.
I think that the only way you can discover if a document is portrait/landscape is to reason with the length of the sides of the bounding box of your outline.
I don't think there's an automatic way to do this, maybe you can find the most external contour approximable with a 4 segment polyline (all doable in opencv). In order to get this you'll have to work with contour hierarchy and contous approximation (see cv2.approxPolyDP).
This is how I would go for automatic outline detection. As I said, the rest of your algorithm seems just fine to me.
PS. I'll leave my Android project GitHub link. I don't know if it can be useful to you, but here I specify the outline by dragging some handles, then transform the image and feed it to Tesseract, using Java and OpenCV. Yeah It's a very bad idea to do that in the main thread of an Android app and yeah, the app is not finished. I just wanted to experiment with OCR, so I didn't care much of performance and usability, since this was not intended to use, but just for studying.
Look up the uniform width transform.
What this does is detect edges which have more or less the same width with respect to their opposite edge. So things like drainpipes (which can be eliminated at a later pass) but also the majority of text. Whilst conceptually it's similar to a distance transform, the published method uses rather ad hoc normal projection methods and Canny edge detection.
I am currently working on an iOS app that can take a picture programmatically using AVFoundation libraries like AVCaptureDevice through a custom button.
The new requirement is that the camera should automatically take a picture when the camera session detects something specific. For example, if the camera is open, and I line up an apple to fill a certain circle part of the capture screen, it should take the picture automatically. We can see this auto capture feature in some banking apps when you submit a mobile check deposit.
Does anyone know of existing libraries(open-source or proprietary) that can analyze images in real time while a user is taking a picture?
The first thing you are going to need to do is decide how you want to detect the apple. You can do this using shape detection, image recognition, or various other methods. This is important because you need to know the approach you want to take before you can identify the best way to implement it.
Once you know how you are going to identify the apple, the easiest way to do real-time image processing like this would be to use an existing augmented reality SDK. For example:
http://www.wikitude.com/products/wikitude-sdk/
http://artoolkit.org/
https://developer.vuforia.com/
If you are feeling really adventurous you could roll your own using AForge or a similar library. I have taken this approach in the past for basic shape detection projects.
Edit
The reason I suggest using an existing AR SDK is because generally they provide a lot of the glue between the camera feed and their API for you and it takes a lot of leg work out of the equation. Even though you won't be using any of the actual "augmentation" part of their SDKs, you can still take advantage of the detection part.
No matter what approach you take, you can think about it in the simplest terms of looking a picture, and figuring out if the item you want is in that picture. How do you decide? In most cases you look for a specific shape or pattern.
I'm developing a app that will showcase products. One of the features of this app is that you will be able to "rotate" the product, using your finger/Pan-Gesture.
I was thinking in implementing this by taking photos of the product from different angles so when you "drag" the image, all I would have to do is switch the image according. If you drag a little, i switch only 1 image... if you drag a lot, i will switch them in cadence making it look like a movie... but i have a concerns and a probable solution:
Is this "performatic"? Since its a art/museum product showcase, the photos will be quite large in size/definition, and loading/switching when "dragged a lot" might be a problem because it would cause "flickering"... And the solution would be: instead of loading pic-by-pic i would put them all inside one massive sheet, and work through them as if they were a sprite...
Is that a good ideia? Or should I stick with the pic-by-pic rotation?
Edit 1: There`s a complicator: the user will be able to zoom in/out and to rotate the product in any axis (X, Y and Z)...
My personal opinion, I don't think this will work the way you hope or the performance and/or aesthetics will not be what you want.
1) Taking individuals shots that you then try to keyframe to based on touch events won't work well because you will have inevitable inconsistencies in 'framing' the shots such that the playback won't be smooth
2) The best way to do this, I suspect, will be to shoot it with video and shoot it with some sort of rig that allows you to keep the camera fixed while rotating the object
3) I'm pretty sure this is how most 'professional' grade product carousel type presentations work
4) Even then you will have more image frames than you need -- not sure whether you plan to embed the images files in app or download on demand -- but that is also a consideration in terms of how much downsampling you'll need to do to reduce frames/file size
Suggestion
Look at shooting these as video (somewhat like described above) and downsampling and removing excess frames using a video editor. Then you could use AVFoundation for playback and use your gestures to 'scrub' into the video frames. I worked on something like this for HTML playback at a large company and I can assure you it was done with video.
Alternatively, if video won't work for you. Your sprite sheet solution might work (consider using SpriteKit). But then keep in mind what I said about trying to keyframe one off camera shots together -- it just won't work well. Maybe a compromise would be to shoot static images but do so by fixing the camera and rotating the objects at very specific increments. That could work as well I suppose but you will need to be very careful about light and other atmospehrics. It doesn't take much variation at all to be detectable to the human eye causing the whole presentation to seem strange. Good luck.
A coder from my company did something like this before using 360 images of an object and it worked just great but it didn't have zoom. Maybe you could add zoom by adding a pinch gesture recognizer and placing the image view into a scroll view to zoom in on the static image.
This scenario sounds like what you really need is a simple 3D model loader library or write it in OpenGL yourself. But this pan and zoom behavior is really basic when you make that jump to 3D so it should be easy to find lots of examples.
All depends on your situation and time constraints :)
I want to manually adjust the white balance using a slider before I start recording video from the camera. I have looked at the AVFoundation Framework but it does not allow to pick a value for WB. What frameworks/classes do I need to to adjust the WB in this way?
I haven't been able to find any info on setting the camera's white balance (though I don't know for sure that it's not possible). But, you can always post-process with the white balance Core Image filter (aka CIWhitePointAdjust).
You can read about applying Core Image filters here.
I want to use an UIImagePicker to have a camera preview being displayed. Over this preview I want to place an overlay view with controls.
Is it possible to apply any effects to the preview which will be displayed from camera? I particularly need to apply a blur effect to the camera preview.
So I want to have a blurred preview from camera and overlay view with controls. If I decide to capture the still image from the camera, I need to have it original without blur effect. So blur effect must applied only to the preview.
Is this possible using such configuration or maybe with AVFoundation being used for accessing the camera preview or maybe somehow else, or that's impossible at all?
With AV foundation you could do almost everything you want since you can obtain single frame from the camera and elaborate them, but it could lead you at a dead-end applying a blur on an image in realtime is a pretty intensive task with laggy video results, that could lead you to waste hours of coding. I would suggest you to use the solution of James WebSster or OpenGL shaders. Take a look at this awesome free library written by one of my favorite guru Brad http://www.sunsetlakesoftware.com/2012/02/12/introducing-gpuimage-framework even if you do not find the right filter, probably it will lead you to a correct implementation of what you want to do.
The right filter is Gaussian blur of course, but I don't know if it is supported, but you could do by yourself.
Almost forgot to say than in iOS 5 you have full access to the Accelerate Framework, made by Apple, you should look also into that.
From the reasonably limited amount of work I've done with UIImagePicker I don't think it is possible to apply the blur to the image you see using programatic filters.
What you might be able to do is to use the overlay to estimate blur. You could do this, for example, by adding an overlay which contains an image of semi-transparent frosted glass.