An image processing method, for receiving an input image and separating pixels having text characteristics and pixels having figure characteristics, includes: applying a first filtering processing for the input image to derive a first image processing result; applying a second filtering processing for the first image processing result to derive a second image processing result, wherein a distribution of filtering parameters of the first filtering processing is different from a distribution of filtering parameters of the second filtering processing; deriving a set of first reference values according to the first image processing result and the second image processing result; and determining whether each pixel within the input image is a text pixel or a figure pixel according to at least the set of the first reference values and a predetermined threshold.