SUMMARYThis paper proposes a method that obtains high-tone images from ultra-low-tone image by using spatial features. Areas are smoothed when the adjacent tone value of the area of interest is continuous, but areas are saved as edges when the adjacent tone value of the area of interest is discontinuous. Thus the area of interest can be classified into six patterns by area. Low-gradation images are changed into high-gradation images by processing with each pattern. Comparative experiments indicated that the proposed method was quantitatively and visually better than existing methods. The proposed method can change ultra-low-tone images into naturally high-gradation images. C⃝ 2015 Wiley Periodicals, Inc. Electron Comm Jpn, 98(7): 53-62, 2015; Published online in Wiley Online Library (wileyonlinelibrary.com).