Face-Detector

Face detection matlab application based on skin tone detection, Binary Masking Maximum and Link Domain Determination.

Theme:

A program that identifies the position of a face from a photo of a person

Goal:

• Exclude other skin effects in the case of human photography
• Aim to correctly position the face

Methods

After loading the image,

1. Skin Detection
2. Binary mask processing
3. Maximum consolidation area estimation
4. Final judgment (aspect ratio and both eyes)

1. Skin Detection

Use the YCbCr color space, which is most used for skin detection Area extraction of skin by threshold

2. Binary mask processing

• Morphology Opening
  • Shrink followed by expansion on an image
  • Imopen function for easy processing


3. Maximum Connectivity Area Estimation

• Maximum Binding Area
• Based on 8 connections (a total of 8 adjacent points including diagonal points), maximum from multiple connection zones
• The bwlabel function takes out the connected area, and the coordinates of the rectangle are obtained for the position of the face.
• The find function can get the index of the maximum concatenated area

4. Final verdict

aspect ratio

• Considering errors such as judging the face along with the neck → Detection range aspect ratio set from 0.5 to 2.0
• For large aspect ratios (greater than 1.6) → height increased by 0.75 times original
• For small aspect ratios (less than 1, 0) → width increased by 0.80 times the original

Eyes

• Establish the presence of the eye by the 8 concatenation judgment method described above
• Before judging, reverse the black and white areas in the range
• Create a findeye.m function and set the return value of 1 and 0 depending on whether both eyes are included

Application execution example

Instruction

English

1. Program main file face_detector.m This program consists of three files: face_detector.m, findeye.m, and skim.m. How to run: Open face_detector.m -> Load any image into I -> Run

2. Application: face_detector.mlapp

   • face_detector.mlapp: The main application file. How to run: Open in MATLAB.

   • face_detector.mlappinstall: The installation program. How to run: Install -> MATLAB App -> Open face_detector in My Apps.

   How to use the application:

   1. Click "Select Image..." to choose a file.
   2. Click "Run" to display the binary image and detection results.
   3. Click "Clear" to reset the program.

日本語

1.プログラム本体 face_detector.m

今回のプログラムは、
face_detector.m
findeye.m
skim.m
の3つのファイルからできている。

実行方法：face_detector.mを開く→任意の画像を読み込んでIに入れる→実行

2.アプリケーション face_detector.mlapp

・face_detector.mlapp:アプリケーション本体
	実行方法：MATLABから開く

・face_detector.mlappinstall:インストールプログラム
	実行方法：インストール→MARTLABのアプリ→マイアプリ→face_detectorを開く

アプリの使い方：
1. 「画像を選択...」ボタンを押し、ファイルを選ぶ。
2. 「実行」ボタンを押し、二値化画像と検出結果が順に示される。
3. 「クリア」ボタンを押し、プログラムがリセットされる。