Analysis of Face Detection by AdaBoost algorithm using Fuzzy Membership

NetBean project:

- fuzzy_gauss_19: implementation of AdaBoost with Gaussian Fuzzy MF on MIT cbcl dataset
    - 20180125_fuzzy_gauss1_19.wk
    - 20180125_fuzzy_gauss2_19.wk
    - 20180125_fuzzy_gauss3_19.wk
- fuzzy_gauss_24: implementation of AdaBoost with Gaussian Fuzzy MF on Yi-qing dataset
    - 20180125_fuzzy_gauss1_24.wk
    - 20180125_fuzzy_gauss3_24.wk
- fuzzy_tri_19: implementation of AdaBoost with Triangular Fuzzy MF on MIT cbcl dataset
    - 20180117_fuzzy_tri_std1_19.wk
    - 20180118_fuzzy_tri_std15_19.wk
    - 20180116_fuzzy_tri_std2_19.wk
    - 20180118_fuzzy_tri_std25_19.wk
    - 20180118_fuzzy_tri_std3_19.wk
- fuzzy_tri_24: implementation of AdaBoost with Triangular Fuzzy MF on Yi-qing dataset
    - 20180119_fuzzy_tri_std1_24.wk
    - 20180119_fuzzy_tri_std15_24.wk
    - 20180117_fuzzy_tri_std2_24.wk
    - 20180119_fuzzy_tri_std25_24.wk
    - 20180119_fuzzy_tri_std3_24.wk
- vj_thres_19: implementation of AdaBoost with Viola-Jones method on MIT cbcl dataset
    - 20171025_vj_5000_19.wk
- vj_thres_24: implementation of AdaBoost with Viola-Jones method on Yi-qing dataset
    - 20171025_vj_24.wk
Each Strongclass (.wk file) mentioned above consists of 200 Weakclass 

Program structure:

main
|__ trainingProcess
|__ testingProcess

Training Process:

trainProcess
|__ readInputImage
|__ calculateIntegralImage
|__ readTypePattern
|__ readIndex
|__ algorithmAdaboost
|__ saveArrWeakClass

Functions:

ArrayList<int[][]> readInputImage(String path)
    Parameter:
        - path: directory path to get input images
    Usage:
        - Read and pre-process input image
    Step:
        - Read images from given path
        - Convert to Grayscale (if using Yi-qing dataset with PNG image)
        - Normalize image
        - Return an ArrayList
ArrayList<int[][]> calculateIntegralImage(ArrayList<int[][]> lstImg)
    Parameter:
        - lstImg: ArrayList of input image
    Usage:
        - Compute integral images
    Step:
        - Get Integral images and store in an ArrayList
ArrayList<int[][]> readTypePattern(String path)
    Parameter:
        - path: directory path of color map
    Usage:
        - Read color map for each Haar-like pattern
IndexObj[] readIndex(String path)
    Parameter:
        - path: directory path of feature index
    Usage:
        - Read all feature index from Lookup table
WeakClass[] algorithmAdaboost()
    Usage:
        - Train Strongclass from given Integral images and Haar-like pattern
        - Each WeakClass is chosen from 5000 random features
    Step:
        - Initialize Weight for Face and Non-face samples
        - For each round:
            - Normalize Weights
            - Get Feature with smallest error
            - Calculate Alpha
            - Update Weights
FeatureIndex getFeatureIndex_weights(double[] weights, int numPosTrain, int numNegTrain)
    Parameter:
        - weights: Weights of all face and non-face samples
        - numPosTrain: number of face samples
        - numNegTrain: number of non-face samples
    Usage:
        - Choose the Feature that yields the smallest error rate with given Weights for Face and Non-face
    Step:
        - Get K (=5000) random features
        - For each feature:
            - Get feature values for the whole set of samples
            - Calculate mean, variance, (standard deviation if Triangular Fuzzy MF)  for each distribution
            - Calculate membership or compare for classification
            - Compute error to choose feature producing least error rate
Boolean isPositiveDist(double meanPos, double varPos, double meanNeg, double varNeg,         double val, double gamma)
    Parameter:
        - meanPos, varPos: mean and variance of Face distribution
        - meanNeg, varNeg: mean and variance of Non-face distribution
    Usage:
        - Compare density function to Face and Non-face distribution without directly calculating the function value
    Step:
        - Just simplify ratio of 2 density function formulas
    Note: only use for Gaussian Fuzzy MF
double getMembershipTriangle(double min, double mean, double max, double val)
    Parametere:
        - min, max: minimum and maximum values of distribution
        - mean: mean of distribution
        - val: value to find the corresponding membership
    Usage:
        - Calculate membership of a value to a Triangular Fuzzy distribution
    Step:
        - Just apply Fuzzy MF formula
    Note: only use for Triangular Fuzzy MF

Test Process:

trainProcess
|__ readInputImage
|__ calculateIntegralImage
|__ readTypePattern
|__ readIndex
|__ loadWeakClass
|__ testData

Functions:

void testData(String fileTestInfo, double rate)
    Parameter:
        - fileTestInfo: file that contains information about the number of face and non-face for testing
        - rate: rate used for testing (instead of using 0.5 as in origin AdaBoost algorihtm)
    Usage:
        - Test StrongClass with given number of WeakClass and rate to find TPR, FPR
    Step:
        - For each test sample:
            - For each Weakclass
                - Calculate feature value
                - Calculate membership to classify
            - Classify face or non-face