This paper introduces an improved preprocessing algorithm to enhance the segmentation accuracy of SLIC. Reducing image resolution and noise is important to fast runtime and high detection accuracy of detection algorithms. However, it causes loss of core image data for detection algorithm, such as edges. Superpixel is a pixel group of similar color. Algorithms making superpixel is a kind of clustering algorithm that reduces noise of images and maintains edge information. Simple Linear Iterative Clustering(SLIC) is a algorithm that more improve both runtime and compactness than the conventional algorithm. But SLIC tends not to exactly express the boundary of image data due to the high compactness. To resolve this problem, we propose a new algorithm that improves the efficiency of segmentation by using gradient information on the basis of SLIC.