Building and feature detection are significant research fields within computer vision, yet differentiating between old and modern architectural structures presents unique computational challenges. While the human eye can easily distinguish between historical and contemporary buildings based on visual cues, automated systems must rely on specific feature extraction methods. This research paper, co-authored by S. Rayhan Kabir, presents a performance analysis of four computational techniques: Canny Edge Detection, Hough Line Transform, Find Contours, and the Harris Corner Detector. The study evaluates these techniques across various modern and old building datasets to determine their effectiveness in identifying unique structural characteristics. By differentiating pixel-level characteristics such as line alignment, edge density, and corner frequency, the experiment demonstrates why these four algorithms are particularly suitable for structural classification. The findings provide a technical foundation for developing automated urban planning tools, heritage preservation systems, and intelligent navigation models that require precise architectural recognition.

Keywords: Computer Vision, Feature Detection, Canny Edge Detection, Harris Corner Detector, Architectural Analysis, Image Processing, AUB.