Posted on

Presenting historical changes of building

The goal of this project is to implement algorithm that extract similar points or whole regions from two different images of the same building using OpenCV library and especially MSER algorithm (Maximally stable extremal regions). Images of building are taken in different time and have different hue, saturation, light and other conditions.

Based on the extracted regions, algorithm finds the same centers of key regions and merged images by these points to create a complete images of building with the presentation of its historical changes.

Functions used: MSER, fitEllipse, adaptiveThreshold, Canny, findContours


The process

  1. Preprocessing
  2. MSER regions detection
    MSER mser(int _delta, int _min_area, int _max_area, float _max_variation, float _min_diversity, int _max_evolution, double _area_threshold, double _min_margin, int _edge_blur_size);

    MSER algorithm with different parameters
  3. Fitting detected regions by ellipse
    const vector<Point>& r;
    RotatedRect box = fitEllipse(r);
  4. Finding similar regions
  5. Merging images based on found regions

Practical Application

Interactive presentation of the historical buildings and visualising their changes in time.