Novel Algorithm for Image Mosaicking using Feature based Corner Detection

A novel feature based corner detection method is proposed to extract accurate corner positions than unstable corner points in existing algorithms such as Intensity-based Harris corner method and Edge-based corner method. The former algorithm cannot detect accurate corner positions in noisy environments while the latter cannot identify stable corners due to incomplete edge detection. The proposed method can overcome limitations of both these methods. This paper proposes a feature based novel algorithm for image mosaicking using block wise corner detection technique and is amenable for hardware realization. The mosaicking algorithm has been coded in MATLAB and successfully tested for color images. The performance metric PSNR has been computed for both the proposed and the existing methods for a number of mosaic images. The proposed method performs better than existing algorithms. The PSNR values obtained for these images are 42 dB or better with no visible artifacts.