ImageBlurDetection Save

To check whether an image is blurred or not in Android, selecting image via Gallery

Project README

ImageBlur Detection

To check whether an image is blurred or not in Android. Image should be selected from Gallery.

OpenCV library is used for Blur detection.

Download OpenCV by selecting a stable version of Android pack from here. Integrating OpenCV to studio-project is quite a long procedure. Please refer this awesome article.

Threshold value

If the sharpness of image is lesser than this value, then that image is marked as blurred.

Screenshots

Blurred Image Not Blurred Image

APK Size

APK size is the main concern when we want to integrate with OpenCV.

Apk size

  • Debug APK size - 86 MB
  • Debug APK size with Progaurd - 84.4 MB
  • Debug APK size with Progaurd and R8 - 80 MB
  • Debug APK size with Progaurd + R8 + NDK abiFilters(including only armv7 and x86) - 24 MB

Workaround to minimise apk size

  • Splitting apks by including only those architectures that we need to support. Also include universal apk.
 splits {
            abi {
                enable true
                reset()
                include "armeabi-v7a","x86"
                universalApk true
            }
        }
  • Creating an apk according to w.r.t architectures included. Also including universal apk.
ext.versionCodes = ['armeabi-v7a': 4, 'x86': 5]
import com.android.build.OutputFile
android.applicationVariants.all { variant ->
    variant.outputs.each { output ->
        def abiFilter = output.getFilter(OutputFile.ABI)
        def abiMultiplier = 0
        if (abiFilter != null) {
            abiMultiplier = project.ext.versionCodes.get(abiFilter)
        }
        output.versionCodeOverride = abiMultiplier * 1000 + android.defaultConfig.versionCode
    }
}

Problem with this approach is we have to upload all the apks separately for google play console, including univeral apk. Credits

Factors which decides the quality of an Image

  • Mega Pixels of Camera - Doesn’t necessarily impact image quality, more MP!=more quality.
  • Camera Sensor quality - Hardware component which decides image quality. Better sensor will result in better picture quality.
  • Pixel Size - more pixel size will give more picture quality. 16MP Samsung S5 and S6 has 1.12µm but 12MP Samsung S7 has 1.4µm.
  • Aperture - Hole/Opening in the Camera through which light travels. Larger apertures are better, To control the amount of light reaching the image sensor. 
  • Shutter Speed - Length of time with which sensor is exposed to light. More shutter speed is, the more light is available to focus.
Ex : For moving objects, slower SP results in blurred images
  • Image Size - Doesn’t necessarily impact image quality, more image size!=more quality.

How to decide threshold value?


Threshold is a critical parameter to verify the blurriness of image.

  • Too small of a value, and you’ll accidentally mark images as blurry when they are not.
  • With too large of a threshold, you’ll mark images as non-blurry when in fact they are. Credits

So finally, What value to use for threshold?

It all depends on the dataset of images. Currently we are keeping it as 500.

Useful info

According to stats, armv7 is the most commonly used android architectures.

Open Source Agenda is not affiliated with "ImageBlurDetection" Project. README Source: jainullas/ImageBlurDetection
Stars
57
Open Issues
3
Last Commit
2 years ago

Open Source Agenda Badge

Open Source Agenda Rating