xumiaomiaomiao/computer-vision_22spring

Computer Vision projects during 22Spring semester at ITMO university

Computer-Vision_22Spring

Computer Vision projects during 22Spring semester at ITMO university

Below is a simple table of contents for the Practice :

• Practice 1 ： basic methods for images segmentation into semantic areas.
  • Binarization.（upper and lower binarization thresholds）
  • Segmentation 1
    • Image segmentation based on Weber principle
    • Segmentation of RGB images by skin color
  • Segmentation 2
    • image segmentation in the CIE Lab color space by the nearest neighbors method .
    • image segmentation in the CIE Lab color space by the 𝑘-means method .
  • Segmentation 3
    • Texture segmentation using mean value, Standard deviation, relative smoothness, local entropy.
• Practice 2 ： Hough Transform.
  • Search for lines.
    • Search for straight lines using the Hough transform both for the original image and for the image obtained using differential operator.
  • Search for circles.
    • Search for circles of both a certain radius and from a given range using the Hough transform, both for the original image and for the image obtained using differential operator.
  • classic Hough transform algorithms for lines, Highlight the selected points in the Hough parameter space.
  • classic Hough transform algorithms for circles, Highlight the selected points in the Hough parameter space.
  • Compare implementation results.
• Practice 3 ： Features Detectors.
  • Feature points detection.
    • Using SIFT feature point descriptor
    • Using ORB feature point descriptor
  • Feature points matching.
    • Extract feature points of an object and match them with feature points of a scene containing this object, Calculate the transformation matrix using RANSAC method and highlight the object position in the scene.
    • Compare feature point descriptors for the task of image matching.
  • simple automatic image stitching.
    • calculate the transformation matrix between two images and stitch them into a single panoramic image.
    • stitch three images into a single panoramic image.
• Practice 4 ： Face Detection using Viola-Jones Approach.
  • Faces detection.
    • search faces using Viola-Jones approach.
  • Body parts detectiong.
    • Search for eyes, mouth, and nose in one image, and use ROI to improve accuracy.
  • face detection in videostream using pre-recorded video with faces.
    • ★bilibili video : Facial Recognition of Fragments from ”Goodluck Charlie ”(Viola-Jones Approach)★
  • face detection in live videostream using web-camera.