Added mathematics library for transformation support. Implemented PHash and added PHash specific tests

pom.xml

@@ -80,6 +80,11 @@
       <artifactId>thumbnailator</artifactId>
       <version>[0.4, 0.5)</version>
     </dependency>
+    <dependency>
+      <groupId>net.sourceforge.jtransforms</groupId>
+      <artifactId>jtransforms</artifactId>
+      <version>2.4.0</version>
+    </dependency>
   </dependencies>
 
   <build>

src/main/resources/default.properties

@@ -24,7 +24,7 @@ image.dhash.tolerence=8
 #set to false if hashing images is taking too long
 image.phash.use=true
 image.phash.weight=1.0
-image.phash.precision=16
+image.phash.precision=32
 image.phash.tolerence=8
 
 #Default Thumbnail Settings

src/main/scala/com/sothr/imagetools/hash/PHash.scala

@@ -1,10 +1,63 @@
 package com.sothr.imagetools.hash
 
+import edu.emory.mathcs.jtransforms.dct.FloatDCT_2D
+import grizzled.slf4j.Logging
+
 /**
   * Created by dev on 1/22/14.
   */
-object PHash extends PerceptualHasher {
+object PHash extends PerceptualHasher with Logging {
    def getHash(imageData: Array[Array[Int]]): Long = {
-     return 0L
+     //convert the imageData into a FloatArray
+     val width = imageData.length
+     val height = imageData(0).length
+
+     val imageDataFloat:Array[Array[Float]] = Array.ofDim[Float](height, width)
+     for (row <- 0 until height) {
+       for (col <- 0 until width) {
+         imageDataFloat(row)(col) = imageData(row)(col).toFloat
+       }
+     }
+
+     //perform transform on the data
+     val dct:FloatDCT_2D = new FloatDCT_2D(height,width)
+     dct.forward(imageDataFloat, true)
+
+     //extract the DCT data
+     val dctDataWidth:Int = width / 4
+     val dctDataHeight:Int = height  / 4
+
+     //calculate the mean
+     var total = 0.0f
+     for (row <- 0 until dctDataHeight) {
+       for (col <- 0 until dctDataWidth) {
+         total += imageDataFloat(row)(col)
+       }
+     }
+     val mean = total / (dctDataHeight * dctDataWidth)
+
+     //calculate the hash
+     var hash = 0L
+     for (row <- 0 until dctDataHeight by 2) {
+       //process each column
+       for (col <- 0 until dctDataWidth by 1) {
+         hash <<= 1
+         val pixel = imageDataFloat(row)(col)
+         //If the current pixel is at or above the mean, store it as a one, else store it as a zero
+         if (pixel >= mean) hash |= 1 else hash |= 0
+       }
+
+       if ((row +1) < dctDataWidth) {
+         val nextRow = row + 1
+         //process each column
+         for (col <- (dctDataWidth - 1) to 0 by -1) {
+           hash <<= 1
+           val pixel = imageDataFloat(nextRow)(col)
+           if (pixel >= mean) hash |= 1 else hash |= 0
+         }
+       }
+     }
+     debug(s"Computed PHash: $hash from ${dctDataWidth * dctDataHeight} pixels")
+     hash
    }
  }

src/test/resources/default.properties

@@ -24,7 +24,7 @@ image.dhash.tolerence=8
 #set to false if hashing images is taking too long
 image.phash.use=true
 image.phash.weight=1.0
-image.phash.precision=16
+image.phash.precision=32
 image.phash.tolerence=8
 
 #Default Thumbnail Settings

src/test/scala/com/sothr/imagetools/hash/HashServiceTest.scala

@@ -101,4 +101,158 @@ class HashServiceTest extends BaseTest {
     assert(HashService.areDhashSimilar(mediumHash,smallHash))
   }
 
+  def ahashTestCase(filePath:String):Long = {
+    val sample = new File(filePath)
+    val image = ImageIO.read(sample)
+    HashService.getAhash(image)
+  }
+
+  test("Benchmark AHash") {
+    info("Benchmarking AHash")
+    info("AHash Large Image 3684x2736")
+    val largeTime1 = getTime { ahashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime2 = getTime { ahashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime3 = getTime { ahashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime4 = getTime { ahashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime5 = getTime { ahashTestCase(TestParams.LargeSampleImage1) }
+    val largeMean = getMean(largeTime1, largeTime2, largeTime3, largeTime4, largeTime5)
+    info(s"The mean time of 5 tests for large was: $largeMean ms")
+    info("AHash Medium Image 1824x1368")
+    val mediumTime1 = getTime { ahashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime2 = getTime { ahashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime3 = getTime { ahashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime4 = getTime { ahashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime5 = getTime { ahashTestCase(TestParams.MediumSampleImage1) }
+    val mediumMean = getMean(mediumTime1, mediumTime2, mediumTime3, mediumTime4, mediumTime5)
+    info(s"The mean time of 5 tests for medium was: $mediumMean ms")
+    info("AHash Small Image 912x684")
+    val smallTime1 = getTime { ahashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime2 = getTime { ahashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime3 = getTime { ahashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime4 = getTime { ahashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime5 = getTime { ahashTestCase(TestParams.SmallSampleImage1) }
+    val smallMean = getMean(smallTime1, smallTime2, smallTime3, smallTime4, smallTime5)
+    info(s"The mean time of 5 tests for small was: $smallMean ms")
+    assert(true)
+  }
+
+  test("Calculate AHash Large Sample Image 1") {
+    debug("Starting 'Calculate AHash Large Sample Image 1' test")
+    val sample = new File(TestParams.LargeSampleImage1)
+    debug(s"Testing File: ${sample.getAbsolutePath} exists: ${sample.exists}")
+    val image = ImageIO.read(sample)
+    debug(s"Image: width: ${image.getWidth} height: ${image.getHeight}")
+    val hash = HashService.getAhash(image)
+    debug(s"Testing that $hash = 36070299219713907L")
+    assert(hash == 36070299219713907L)
+  }
+
+  test("Calculate AHash Medium Sample Image 1") {
+    debug("Starting 'Calculate AHash Medium Sample Image 1' test")
+    val sample = new File(TestParams.MediumSampleImage1)
+    debug(s"Testing File: ${sample.getAbsolutePath} exists: ${sample.exists}")
+    val image = ImageIO.read(sample)
+    debug(s"Image: width: ${image.getWidth} height: ${image.getHeight}")
+    val hash = HashService.getAhash(image)
+    debug(s"Testing that $hash = 36070299219713907L")
+    assert(hash == 36070299219713907L)
+  }
+
+  test("Calculate AHash Small Sample Image 1") {
+    debug("Starting 'Calculate AHash Small Sample Image 1' test")
+    val sample = new File(TestParams.SmallSampleImage1)
+    debug(s"Testing File: ${sample.getAbsolutePath} exists: ${sample.exists}")
+    val image = ImageIO.read(sample)
+    debug(s"Image: width: ${image.getWidth} height: ${image.getHeight}")
+    val hash = HashService.getAhash(image)
+    debug(s"Testing that $hash = 36070299219713907L")
+    assert(hash == 36070299219713907L)
+  }
+
+  test("AHash Of Large, Medium, And Small Sample 1 Must Be Similar") {
+    val largeHash = ahashTestCase(TestParams.LargeSampleImage1)
+    val mediumHash = ahashTestCase(TestParams.MediumSampleImage1)
+    val smallHash = ahashTestCase(TestParams.SmallSampleImage1)
+    assert(HashService.areAhashSimilar(largeHash,mediumHash))
+    assert(HashService.areAhashSimilar(largeHash,smallHash))
+    assert(HashService.areAhashSimilar(mediumHash,smallHash))
+  }
+
+  def phashTestCase(filePath:String):Long = {
+    val sample = new File(filePath)
+    val image = ImageIO.read(sample)
+    HashService.getPhash(image)
+  }
+
+  test("Benchmark PHash") {
+    info("Benchmarking PHash")
+    info("PHash Large Image 3684x2736")
+    val largeTime1 = getTime { phashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime2 = getTime { phashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime3 = getTime { phashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime4 = getTime { phashTestCase(TestParams.LargeSampleImage1) }
+    val largeTime5 = getTime { phashTestCase(TestParams.LargeSampleImage1) }
+    val largeMean = getMean(largeTime1, largeTime2, largeTime3, largeTime4, largeTime5)
+    info(s"The mean time of 5 tests for large was: $largeMean ms")
+    info("PHash Medium Image 1824x1368")
+    val mediumTime1 = getTime { phashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime2 = getTime { phashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime3 = getTime { phashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime4 = getTime { phashTestCase(TestParams.MediumSampleImage1) }
+    val mediumTime5 = getTime { phashTestCase(TestParams.MediumSampleImage1) }
+    val mediumMean = getMean(mediumTime1, mediumTime2, mediumTime3, mediumTime4, mediumTime5)
+    info(s"The mean time of 5 tests for medium was: $mediumMean ms")
+    info("PHash Small Image 912x684")
+    val smallTime1 = getTime { phashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime2 = getTime { phashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime3 = getTime { phashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime4 = getTime { phashTestCase(TestParams.SmallSampleImage1) }
+    val smallTime5 = getTime { phashTestCase(TestParams.SmallSampleImage1) }
+    val smallMean = getMean(smallTime1, smallTime2, smallTime3, smallTime4, smallTime5)
+    info(s"The mean time of 5 tests for small was: $smallMean ms")
+    assert(true)
+  }
+
+  test("Calculate PHash Large Sample Image 1") {
+    debug("Starting 'Calculate PHash Large Sample Image 1' test")
+    val sample = new File(TestParams.LargeSampleImage1)
+    debug(s"Testing File: ${sample.getAbsolutePath} exists: ${sample.exists}")
+    val image = ImageIO.read(sample)
+    debug(s"Image: width: ${image.getWidth} height: ${image.getHeight}")
+    val hash = HashService.getPhash(image)
+    debug(s"Testing that $hash = -9154589787976242949L")
+    assert(hash == -9154589787976242949L)
+  }
+
+  test("Calculate PHash Medium Sample Image 1") {
+    debug("Starting 'Calculate PHash Medium Sample Image 1' test")
+    val sample = new File(TestParams.MediumSampleImage1)
+    debug(s"Testing File: ${sample.getAbsolutePath} exists: ${sample.exists}")
+    val image = ImageIO.read(sample)
+    debug(s"Image: width: ${image.getWidth} height: ${image.getHeight}")
+    val hash = HashService.getPhash(image)
+    debug(s"Testing that $hash = -9154589787976242949L")
+    assert(hash == -9154589787976242949L)
+  }
+
+  test("Calculate PHash Small Sample Image 1") {
+    debug("Starting 'Calculate PHash Small Sample Image 1' test")
+    val sample = new File(TestParams.SmallSampleImage1)
+    debug(s"Testing File: ${sample.getAbsolutePath} exists: ${sample.exists}")
+    val image = ImageIO.read(sample)
+    debug(s"Image: width: ${image.getWidth} height: ${image.getHeight}")
+    val hash = HashService.getPhash(image)
+    debug(s"Testing that $hash = -9154589787976242949L")
+    assert(hash == -9154589787976242949L)
+  }
+
+  test("PHash Of Large, Medium, And Small Sample 1 Must Be Similar") {
+    val largeHash = phashTestCase(TestParams.LargeSampleImage1)
+    val mediumHash = phashTestCase(TestParams.MediumSampleImage1)
+    val smallHash = phashTestCase(TestParams.SmallSampleImage1)
+    assert(HashService.arePhashSimilar(largeHash,mediumHash))
+    assert(HashService.arePhashSimilar(largeHash,smallHash))
+    assert(HashService.arePhashSimilar(mediumHash,smallHash))
+  }
+
 }