This repository contains the final project for the Digital Image Processing course, taken during the 8th semester at EECE, DUTh. The purpose of this project was to evaluate whether the Template Matching method works more effectively on colored images (in RGB or YCbCr formats) or on grayscale images.
The repository includes the following MATLAB scripts:
script_rgb.m: Applies the template matching method on RGB and grayscale images.script_ycbcr.m: Applies the template matching method on YCbCr format images with varying weights on luma and chroma components.Temp_Match_RGB.m: Contains the core function for performing template matching on RGB images.draw_match.m: Visualizes the matching results on the images.local_max.m: Finds local maxima in the similarity map for template matching.
For the selected test images, the following results were observed:
- Grayscale Images: The algorithm can detect the smiley object in the desired regions, but it sometimes identifies false positives where the object does not exist.
- RGB Images: The algorithm performs better in detecting the smiley object, especially when using the mean method. However, the weight method does not yield as accurate results.
- Higher Weight on Luma Component (ycbcr2): The algorithm struggles to detect the object effectively.
- Higher Weight on Chroma Components (ycbcr3): The algorithm performs as well as in the RGB format, showing successful template matching.
The project demonstrated that template matching yields better results on RGB images compared to grayscale images, particularly when using the mean method. When applied to YCbCr format images, the algorithm performed well when the chroma components were given more weight than the luma component.