Tutorial: Image processing on iOS

Table of Contents

• Introduction
• StartedImageProc application
• Image conversion functions
  • UIImage to color vpImage
  • UIImage to gray vpImage
  • Color vpImage to UIImage
  • Gray vpImage to UIImage
• Application output
• Known issues
  • libxml/parser.h file not found
• Next tutorial

This tutorial supposes that you have followed the Tutorial: How to create a basic iOS application that uses ViSP.

Introduction

In this tutorial you will learn how to do simple image processing on iOS devices with ViSP. This application loads a color image (monkey.png) and allows the user to visualize either this image in grey level, either the image gradients, or either canny edges on iOS simulator or devices.

In ViSP images are carried out using vpImage class. However in iOS, image rendering has to be done using UIImage class that is part of the Core Graphics framework available in iOS. In this tutorial we provide the functions that allow to convert a vpImage to an UIImage and vice versa.

Note that all the material (source code and image) used in this tutorial is part of ViSP source code and could be downloaded using the following command:

$ svn export https://github.com/lagadic/visp.git/trunk/tutorial/ios/StartedImageProc

StartedImageProc application

Let us consider the Xcode project named StartedImageProc that is part of ViSP source code. This project is a Xcode "Single view application" where we renamed ViewController.m into ViewController.mm, introduced minor modifications in ViewController.h and add monkey.png image.

To download the complete StartedImageProc project, run the following command:

$ svn export https://github.com/lagadic/visp.git/trunk/tutorial/ios/StartedImageProc

Once downloaded, you have just to drag & drop ViSP and OpenCV frameworks available following Tutorial: Installation from prebuilt packages for iOS devices.

In the dialog box, enable check box "Copy item if needed" to add visp3.framework and opencv.framework to the project.

Now you should be able to build and run your application.

Image conversion functions

The Xcode project StartedImageProc contains ImageConversion.h and ImageConversion.mm files that implement the functions to convert UIImage to ViSP vpImage and vice versa.

UIImage to color vpImage

The following function implemented in ImageConversion.mm show how to convert an UIImage into a vpImage<vpRGBa> instanciated as a color image.

// Converts an UIImage that could be in gray or color into a ViSP color image
+ (vpImage<vpRGBa>)vpImageColorFromUIImage:(UIImage *)image
{
  CGColorSpaceRef colorSpace = CGImageGetColorSpace(image.CGImage);
  
  if (CGColorSpaceGetModel(colorSpace) == kCGColorSpaceModelMonochrome) {
    NSLog(@"Input UIImage is grayscale");
    vpImage<unsigned char> gray(image.size.height, image.size.width); // 8 bits per component, 1 channel
    
    CGContextRef contextRef = CGBitmapContextCreate(gray.bitmap,                // pointer to  data
                                                    image.size.width,           // width of bitmap
                                                    image.size.height,          // height of bitmap
                                                    8,                          // bits per component
                                                    image.size.width,           // bytes per row
                                                    colorSpace,                 // colorspace
                                                    kCGImageAlphaNone |
                                                    kCGBitmapByteOrderDefault); // bitmap info flags
    
    CGContextDrawImage(contextRef, CGRectMake(0, 0, image.size.width, image.size.height), image.CGImage);
    CGContextRelease(contextRef);
    
    vpImage<vpRGBa> color;
    vpImageConvert::convert(gray, color);
    
    return color;
  }
  else {
    NSLog(@"Input UIImage is color");
    vpImage<vpRGBa> color(image.size.height, image.size.width); // 8 bits per component, 4 channels
    
    colorSpace = CGColorSpaceCreateDeviceRGB();
    
    CGContextRef contextRef = CGBitmapContextCreate(color.bitmap,               // pointer to  data
                                                    image.size.width,           // width of bitmap
                                                    image.size.height,          // height of bitmap
                                                    8,                          // bits per component
                                                    4 * image.size.width,       // bytes per row
                                                    colorSpace,                 // colorspace
                                                    kCGImageAlphaNoneSkipLast |
                                                    kCGBitmapByteOrderDefault); // bitmap info flags
    
    CGContextDrawImage(contextRef, CGRectMake(0, 0, image.size.width, image.size.height), image.CGImage);
    CGContextRelease(contextRef);
    
    return color;
  }
}

UIImage to gray vpImage

The following function implemented in ImageConversion.mm show how to convert an UIImage into a vpImage<unsigned char> instanciated as a grey level image.

// Converts an UIImage that could be in gray or color into a ViSP gray image
+ (vpImage<unsigned char>)vpImageGrayFromUIImage:(UIImage *)image
{
  CGColorSpaceRef colorSpace = CGImageGetColorSpace(image.CGImage);
  
  if (CGColorSpaceGetModel(colorSpace) == kCGColorSpaceModelMonochrome) {
    NSLog(@"Input UIImage is grayscale");
    vpImage<unsigned char> gray(image.size.height, image.size.width); // 8 bits per component, 1 channel
    
    CGContextRef contextRef = CGBitmapContextCreate(gray.bitmap,                // pointer to  data
                                                    image.size.width,           // width of bitmap
                                                    image.size.height,          // height of bitmap
                                                    8,                          // bits per component
                                                    image.size.width,           // bytes per row
                                                    colorSpace,                 // colorspace
                                                    kCGImageAlphaNone |
                                                    kCGBitmapByteOrderDefault); // bitmap info flags
    
    CGContextDrawImage(contextRef, CGRectMake(0, 0, image.size.width, image.size.height), image.CGImage);
    CGContextRelease(contextRef);
    
    return gray;
  } else {
    NSLog(@"Input UIImage is color");
    vpImage<vpRGBa> color(image.size.height, image.size.width); // 8 bits per component, 4 channels (color channels + alpha)
    
    colorSpace = CGColorSpaceCreateDeviceRGB();
    
    CGContextRef contextRef = CGBitmapContextCreate(color.bitmap,               // pointer to  data
                                                    image.size.width,           // width of bitmap
                                                    image.size.height,          // height of bitmap
                                                    8,                          // bits per component
                                                    4 * image.size.width,       // bytes per row
                                                    colorSpace,                 // colorspace
                                                    kCGImageAlphaNoneSkipLast |
                                                    kCGBitmapByteOrderDefault); // bitmap info flags
    
    CGContextDrawImage(contextRef, CGRectMake(0, 0, image.size.width, image.size.height), image.CGImage);
    CGContextRelease(contextRef);
    
    vpImage<unsigned char> gray;
    vpImageConvert::convert(color, gray);
    
    return gray;
  }
}

Color vpImage to UIImage

The following function implemented in ImageConversion.mm show how to convert a gray level vpImage<unsigned char> into an UIImage.

// Converts a color ViSP image into a color UIImage
+ (UIImage *)UIImageFromVpImageColor:(const vpImage<vpRGBa> &)I
{
  NSData *data = [NSData dataWithBytes:I.bitmap length:I.getSize()*4];
  CGColorSpaceRef colorSpace;
  
  colorSpace = CGColorSpaceCreateDeviceRGB();
  
  CGDataProviderRef provider = CGDataProviderCreateWithCFData((__bridge CFDataRef)data);
  
  // Creating CGImage from vpImage
  CGImageRef imageRef = CGImageCreate(I.getWidth(),                               // width
                                      I.getHeight(),                              // height
                                      8,                                          // bits per component
                                      8 * 4,                                      // bits per pixel
                                      4 * I.getWidth(),                           // bytesPerRow
                                      colorSpace,                                 // colorspace
                                      kCGImageAlphaNone|kCGBitmapByteOrderDefault,// bitmap info
                                      provider,                                   // CGDataProviderRef
                                      NULL,                                       // decode
                                      false,                                      // should interpolate
                                      kCGRenderingIntentDefault                   // intent
                                      );
  
  
  // Getting UIImage from CGImage
  UIImage *finalImage = [UIImage imageWithCGImage:imageRef];
  CGImageRelease(imageRef);
  CGDataProviderRelease(provider);
  CGColorSpaceRelease(colorSpace);
  
  return finalImage;
}

Gray vpImage to UIImage

The following function implemented in ImageConversion.mm show how to convert a color vpImage<vpRGBa> into an UIImage.

// Converts a gray level ViSP image into a gray level UIImage
+ (UIImage *)UIImageFromVpImageGray:(const vpImage<unsigned char> &)I
{
  NSData *data = [NSData dataWithBytes:I.bitmap length:I.getSize()];
  CGColorSpaceRef colorSpace;
  
  colorSpace = CGColorSpaceCreateDeviceGray();
  
  CGDataProviderRef provider = CGDataProviderCreateWithCFData((__bridge CFDataRef)data);
  
  // Creating CGImage from vpImage
  CGImageRef imageRef = CGImageCreate(I.getWidth(),                               // width
                                      I.getHeight(),                              // height
                                      8,                                          // bits per component
                                      8,                                          // bits per pixel
                                      I.getWidth(),                               // bytesPerRow
                                      colorSpace,                                 // colorspace
                                      kCGImageAlphaNone|kCGBitmapByteOrderDefault,// bitmap info
                                      provider,                                   // CGDataProviderRef
                                      NULL,                                       // decode
                                      false,                                      // should interpolate
                                      kCGRenderingIntentDefault                   // intent
                                      );
  
  
  // Getting UIImage from CGImage
  UIImage *finalImage = [UIImage imageWithCGImage:imageRef];
  CGImageRelease(imageRef);
  CGDataProviderRelease(provider);
  CGColorSpaceRelease(colorSpace);
  
  return finalImage;
}

Application output

• Now we are ready to build "StartedImaeProc" application using Xcode "Product>Build" menu. If you get build issues, have a look to Known issues.
• Once build, if you run StartedImageProc application on your device, you should be able to see the following screen shots.
• Selecting "load image" gives the following result:
  
• Selecting "convert to gray" gives the following result:
  
• Selecting "compute gradient" gives the following result:
  
• Selecting "canny detector" gives the following result:
  

Known issues

libxml/parser.h file not found

When building the project, if you get the following issue libxml/parser.h file not found while including vpXmlParser.h it means that visp3.framework was build with libxml2 third-party, but that the libxml2 headers are not found.

In Xcode, the fix consists in modifying the Build Settings in order to add the following to your Header Search Paths:

$SDKROOT/usr/include/libxml2

This is done in the next screenshot:

Next tutorial

You are now ready to see the Tutorial: AprilTag marker detection on iOS.