In this paper, we propose a novel image scaling algorithm based on the multichannel sampling in the linear canonical transform domain. In spite of the original image, we also consider the first and second derivatives of an image to improve the quality of the reshaped image. The proposed algorithm utilizes the second derivative sampling formula skilfully derived in the paper. In order to verify the effectiveness of the proposed algorithm, we compare it with three image scaling algorithms, i.e., the nearest neighbor algorithm, the bilinear algorithm and the Papoulis algorithm. Experimental results show the proposed algorithm yields significant improvements over the other three algorithms in performance according to different evaluation criteria.