GPUImage2
GPUImage2 copied to clipboard
BradLarson
Hough Transform for Line Detection
It was in the first GPUImage but hasn't been ported over yet. This would also require the Parallel Coordinates transform be ported over.
started work on this here: https://github.com/maxcan/GPUImage2/tree/mc01.1-add-hough-line-detection
very preliminary, but any thoughts / feedback / suggestions welcome.
Im working on an archery scoring program, and I'd LOVE to see if I can test this. I'm a complete image processing and iOS/Mac newb, so any help you can give to me to get this up and running on a test image would be great.
This would be awesome to see. We're working on a screen-based OCR system, this would be a key piece in detecting lines of text! (Combined with being able to attach GPUImage to a screen recording AVCaptureSession).
I'm working on driver license detection and recognition for OCR system. I'm capturing card image using AVcapturesession and trying rectangle detection by using Hough transform. So I need preprocess captured image by using GPUImage2 library in Swift 5.0, but I can't build it in Swift 5.0. there are so many errors.
Ase, not much help on using Hough for this, but there's probably other solutions as there are a lot of CC capture apps right? EG -- they're just scanning the barcode!
https://scannrapp.com
W
On Thu, Apr 11, 2019 at 7:48 AM Ase Naritoshi [email protected] wrote:
I'm working on driver license detection and recognition for OCR system. I'm capturing card image using AVcapturesession and trying rectangle detection by using Hough transform. So I need preprocess captured image by using GPUImage2 library in Swift 5.0, but I can't build it in Swift 5.0. there are so many errors.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/BradLarson/GPUImage2/issues/50#issuecomment-482145926, or mute the thread https://github.com/notifications/unsubscribe-auth/AGP1RnruFRnFZx2Mggsi3Eiykmtty787ks5vf0s0gaJpZM4JY9Jf .
@Ase, I found this that might help you? https://github.com/garnele007/SwiftOCR no experience, but I'd have started with a higher level library. I'm lurking here as I'm hoping one day to detect arrow shafts in a target, which is pretty niche, but it feels like card scanning is a "solved" problem at this point, so unless you're doing it for a learning project, I'd just find something that's targeted at your problem.
On Thu, Apr 11, 2019 at 7:48 AM Ase Naritoshi [email protected] wrote:
I'm working on driver license detection and recognition for OCR system. I'm capturing card image using AVcapturesession and trying rectangle detection by using Hough transform. So I need preprocess captured image by using GPUImage2 library in Swift 5.0, but I can't build it in Swift 5.0. there are so many errors.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/BradLarson/GPUImage2/issues/50#issuecomment-482145926, or mute the thread https://github.com/notifications/unsubscribe-auth/AGP1RnruFRnFZx2Mggsi3Eiykmtty787ks5vf0s0gaJpZM4JY9Jf .