技術(code)參考

多重選取圖片:

使用imagecursor去記憶體拿照片，還要對應顯示圖片位置。

Cursor imagecursor = managedQuery(

MediaStore.Images.Media.EXTERNAL_CONTENT_URI, columns, null,null, orderBy + "DESC");

galleryImageUrls = new ArrayList();

for (int i = 0; i < imagecursor.getCount(); i++) {

imagecursor.moveToPosition(i);

int dataColumnIndex =imagecursor.getColumnIndex(MediaStore.Images.Media.DATA); galleryImageUrls.add(imagecursor.getString(dataColumnIndex)); }

人臉辨識:

使用android內建face detection功能，及可計算出人臉座標、人眼中點座標和偵測到的人臉數。但是手機自動存取圖片時是存放成橫的格式，因此需要把照片轉90度再做人臉辨識。

m.setRotate(-90);

bm = Bitmap.createBitmap(bm, 0, 0, imageWidth, imageHeight, m, true);

Bitmap tempbitmap = Bitmap.createBitmap(bm.getWidth(), bm.getHeight(), Bitmap.Config.RGB_565);

myFace = new FaceDetector.Face[numberOfFace];

myFaceDetect = new FaceDetector(bm.getWidth(), bm.getHeight(), numberOfFace);

numberOfFaceDetected = myFaceDetect.findFaces(bm, myFace);

binarymap = bm.copy(Bitmap.Config.ARGB_8888, true);

眼高辨識:

做眼高辨識時要先將圖片二值化，用偵測出的人眼座標為基準，往上(下)移動尋找，遇到連續兩個白點即便是為眼睛上緣(下)。

for (int i = 0; i < bm.getWidth(); i++) {

for (int j = 0; j < bm.getHeight(); j++) {

int col = binarymap.getPixel(i, j);

int alpha = col & 0xFF000000;

int red = (col & 0x00FF0000) >> 16;

int green = (col & 0x0000FF00) >> 8;

int blue = (col & 0x000000FF);

int gray = (int) ((float) red * 0.3 + (float) green * 0.59 + (float) blue * 0.11);

if (gray <= 100)

gray = 0;

else

gray = 255;

int newColor = alpha | (gray << 16) | (gray << 8) | gray;

binarymap.setPixel(i, j, newColor);

}

}

for (int a = (int) y; a < (int) (y + (myEyesDistance / 2.0)); a++) {

int col = binarymap.getPixel((int) (x - myEyesDistance / 2), a);

int red = (col & 0x00FF0000) >> 16;

int green = (col & 0x0000FF00) >> 8;

int blue = (col & 0x000000FF);

int gray = (int) ((float) red * 0.3 + (float) green * 0.59 + (float) blue * 0.11);

if (gray == 255)

counter1++;

if (counter1 > 2) {

up[k] = a;

canvas.drawLine(x - myEyesDistance / 2, up[k], x + myEyesDistance / 2, up[k], myPaint1);

counter1 = 0;

a = (int) (y + (myEyesDistance / 2));

}

}

for (int a = (int) y; a > (int) (y - (myEyesDistance / 2)); a--) {

int col = binarymap.getPixel((int) (x - myEyesDistance / 2), a);

int red = (col & 0x00FF0000) >> 16;

int green = (col & 0x0000FF00) >> 8;

int blue = (col & 0x000000FF);

int gray = (int) ((float) red * 0.3 + (float) green * 0.59 + (float) blue * 0.11);

if (gray == 255) {

counter++;

}

if (counter > 2) {

down[k] = a;

canvas.drawLine(x - myEyesDistance / 2, down[k], x + myEyesDistance / 2, down[k], myPaint1);

counter = 0;

a = (int) (y - (myEyesDistance / 2));

}

}

height[k] = up[k] - down[k]; }

比較和裁減:

比較:比較個人在十連拍中眼睛睜最大的一張照片。

if (max < eye_h.get(i)) {

max_i = i;

max = eye_h.get(i);

}

裁剪:訂定一張照片為底圖，將每個人的最佳臉貼上合成，使用getpixel一個一個pixel慢慢合成會造成速度過慢，因為太多資料需要處理，因此改用getpixels直接剪下一塊資料在貼上，可以加快速度，在十秒內可完成APP的計算。

BmCrop.getPixels(pixels, 0, BmCrop.getWidth(),xPixel,yPixel, width, heigh); mBitmap.setPixels(pixels, 0, mBitmap.getWidth(),xPixel,yPixel, width, heigh);