Scribd Logo
Upload

Welcome to Scribd!

  • 42 views

    Hand Gesturevrecognization Seminar

    Uploaded by

    AishwaryaSH
    The document discusses hand gesture recognition based on convexity approach and background subtraction. It describes background subtraction techniques like frame differencing and Gaussian mixture models to extract the hand from the background. It then explains the convexity approach which involves extracting the hand contour, approximating it with a polygon, detecting convex and concave points to filter them. Features like convexity, fingertip positions and relative angles are then extracted for classification.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd

    Hand Gesturevrecognization Seminar

    Uploaded by

    AishwaryaSH
    42 views37 pages
    The document discusses hand gesture recognition based on convexity approach and background subtraction. It describes background subtraction techniques like frame differencing and Gaussian mixture models to extract the hand from the background. It then explains the convexity approach which involves extracting the hand contour, approximating it with a polygon, detecting convex and concave points to filter them. Features like convexity, fingertip positions and relative angles are then extracted for classification.

    Original Description:

    Power point presentation on hand gesture recognization

    Copyright

    © © All Rights Reserved

    Available Formats

    PPTX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses hand gesture recognition based on convexity approach and background subtraction. It describes background subtraction techniques like frame differencing and Gaussian mixture models to extract the hand from the background. It then explains the convexity approach which involves extracting the hand contour, approximating it with a polygon, detecting convex and concave points to filter them. Features like convexity, fingertip positions and relative angles are then extracted for classification.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    42 views37 pages

    Hand Gesturevrecognization Seminar

    Uploaded by

    AishwaryaSH
    The document discusses hand gesture recognition based on convexity approach and background subtraction. It describes background subtraction techniques like frame differencing and Gaussian mixture models to extract the hand from the background. It then explains the convexity approach which involves extracting the hand contour, approximating it with a polygon, detecting convex and concave points to filter them. Features like convexity, fingertip positions and relative angles are then extracted for classification.

    Copyright:

    © All Rights Reserved
    Download as PPTX, PDF, TXT or read online from Scribd
    Download as pptx, pdf, or txt
    You are on page 1of 37

    HAND GESTURE RECOGNITION BASED

    ON CONVEXITY APPROACH AND


    BACKGROUND SUBTRACTION
    BY: AISHWARYA S. HUNDEKAR
    2KL18SCS01
    WHAT IS GESTURE?
     Most natural mode of communication

     Gestures may be specified as the physical actions wrought


    by humans that transmit some meaningful information.

     In Human Computer Interaction (HCI), using gestures


    communication is simple and natural as compare to
    graphical user interface as keyboards, computer mice, and
    touch screens.
    GENERALLY, THE PROCESS OF HAND GESTURE
    RECOGNIZATION CAN BE DIVIDED INTO FOUR STAGES

    sample image identification


    feature parameters
    images pre- analysis extraction,
    capturing processing classification and
    recognition
    BACKGROUND SUBTRACTION
     Background subtraction is a common and widely used
    technique for generating a foreground mask by using static
    cameras.
     BS calculates the foreground mask performing a subtraction
    between the current frame and a background model,
    containing the static part of the scene.
     Fundamental logic for detecting moving objects from the
    difference between the current frame and a reference
    frame, called “background image” and this method is known
    as FRAME DIFFERENCE METHOD.
     Background is estimated to be the previous frame.
    Background subtraction equation then becomes:
    B(x ,y ,t) = I (x ,y ,t-1)

    I (x, y, t) – I (x, y, t – 1) > Th



    THRESHOLDING:

    • The key parameter in the thresholding process is the choice of the threshold
    value.
    • Several different methods for choosing a threshold exist,
    – Users can manually choose a threshold value, or
    – A thresholding algorithm can compute a value automatically, which is known as
    automatic thresholding.
    FRAME DIFFERENCING
    INTRODUCTION TO GAUSSIANS:
    • Each pixel is characterized by its intensity in the RGB color space.
    • The probability of observing the current pixel value is considered given by
    the following formula

    • Gaussian probability density function:

    K determined by the available memory and computational power. Stauffer and Grimson
    proposed to set K from 3 to 5
    MIXTURE OF GAUSSIANS:

    • Alternate method for background substraction.


    • the MoG method tracks multiple Gaussian distributions simultaneously.
    • MoG maintains a density function for each pixel.
    • Since MoG is parametric, the model parameters can be adaptively updated without
    keeping a large buffer of video frames.
    • the pixel posses of a pixel Z
    that is the set of the last t values at the pixel ie
    {Z0 , Z1 , ..., Zt−1}
    GAUSSIAN MIXTURE MODEL (GMM)

    • A GMM is a mixture pdf which is a linear


    combination of K Gaussian pdfs.

    • sum of wi = 1

    • each pixel is given one GMM


    CONVEXITY APPROACH:

    • The convexity detection comprises the following steps:


    1. filtering the depth image using erode and dilate approaches,
    2. extracting hand contours,
    3. approximating the hand contour with a polygon,
    4. detecting concave and convex points of the approximation polygon,
    5. Then filtering convex and concave point.
    DILATION:

    • In dilation, a small image called structuring element is used as a local


    maximum operator. As the structuring element is scanned over the image, we
    compute the maximal pixel value overlapped by SE and replace the image
    pixel under the anchor point with that maximal value.

    • Dilation gradually enlarges the boundaries of regions of foreground pixels.


    Thus areas of foreground regions grow in size while holes within those regions
    become smaller.
    EXAMPLE FOR DILATION:
    EROSION:

    • Erosion is the converse of dilation.


    • The action of the erosion operator is equivalent to computing a local minimum
    over the area of the kernel.
    • As the kernel is scanned over the image, we compute the minimal pixel value
    overlapped by SE and replace the image pixel under the anchor point with
    that minimal value.
    EXAMPLE FOR EROSION
    EXTRACTING HAND CONTOUR:

    • The contour of hand is a series of points which are the boundary pixels of the
    hand area.
    • After obtaining the contour, the gesture and its
    shape then can be detected and recognized by
    using contours analysis
    Here we use Canny edge detection method to
    extract the contour of hand gesture
    CANNY EDGE DETECTION ALGORITHM:

    The Canny edge detection algorithm is composed of 5 steps:


    • Noise reduction;
    • Gradient calculation;
    • Non-maximum suppression;
    • Double threshold;
    • Edge Tracking by Hysteresis.
    NOISE REDUCTION:

    • Edge detection results are highly sensitive to image noise.


    • Apply Gaussian blur to smoothen the image.
    • To do so, image convolution technique is applied with a Gaussian Kernel.
    • The kernel size depends on the expected blurring effect. Basically, the
    smallest the kernel, the less visible is the blur.
    GRADIENT CALCULATION:
    • The Gradient calculation step detects the edge intensity and direction by
    calculating the gradient of the image.
    • Edges correspond to a change of pixels’ intensity.
    • To detect it, the easiest way is to apply filters that highlight this intensity
    change in both directions: horizontal (x) and vertical (y)
    • When the image is smoothed, the derivatives Ix and Iy w.r.t. x and y are
    calculated.
    • It can be implemented by convolving I with Sobel kernels Kx and Ky,
    respectively;

    • Then, the magnitude G and the slope θ of the gradient are calculated as
    follow:
    NON-MAXIMUM SUPPRESSION

    • Ideally, the final image should have thin edges.


    • Thus, we must perform non-maximum suppression to thin out the edges.
    • The principle is simple: the algorithm goes through all the points on the
    gradient intensity matrix and finds the pixels with the maximum value in the
    edge directions.
    DOUBLE THRESHOLD:
    EDGE TRACKING BY HYSTERESIS:
    CONVEX HULL:

    • The convex hull of hand gesture contour is the convex polygon surrounded by
    all the convex vertices in gesture contour,
    CONVEXITY DEFECTS:

    • The convex defect is defined as the difference between gesture convex hull and
    contour.
    • The convex defect is defined as the difference
    between gesture convex hull and contour.
    • The data structure of each of the convex
    defects contains three components:
    start contour point, end contour point and
    concave contour point.
    • Therefore, it is possible to detect the fingertips by using hand gesture contour
    and convex defects.
    • The count and position of fingertips can be determined as following:
    1) Conduct noise elimination on the obtained convex defects.
    Based on the physiological structure of human hand and large numbers of experiments,

    it is appropriate to define the depth minimum and maximum threshold as big as

    the 1/5 and 1/2 of the height of hand gesture contour respectively, and the height of

    hand gesture contour is defined as the length of the quadrilateral formed by points

    composed of the co-ordinates.

    2) Scan filtered convex defects clockwise, take the start contour point of the first convex defect and the end contour point
    of the last convex defect as the first and last fingertip respectively.

    3) Take use of the average position of the end contour point of current convex defect and the start contour point of next
    convex defect as the position of current fingertip.
    FEATURES EXTRACTION
    • The tightness of hand gesture contour to its convex hull is defined as the gesture
    convexity, which is denoted by δ, its value is the area ratio of gesture contour and
    convex hull.

    • Where, hullArea is the area of convex hull, contourArea is the area of gesture
    contour.
    • we can get according to the image that hullArea > contourArea, so δ ϵ (0,1), and the
    bigger the value, the tighter the gesture contour to convex hull.
    RELATIVE ANGLES OF FINGERS:
    • Different gestures can be distinguished by the relative position of fingertips,
    which is composed of two values of α and β.
    • α is the summation of angles with centroid as vertex, lines from centroid to the
    first fingertip and other fingertips as edges, and β is the value of the angle
    with centroid as vertex, lines from centroid to the first fingertip and last
    fingertip as edges
    RESULTS AND CONCLUSION

    You might also like