Digital images may register states of a process, and image analysis help us to study this process and its peculiarities. In spite of the fact that there are a lot of textbooks on techniques of image analysis, any new problem may require a new method. Last decades fractal and multifractal methods are often used to analyze high resolution images having complex textures. In many cases fractal characteristics may be considered as classifying signs for further clustering. In this paper we present two multifractal methods. The first method allows us to obtain multifractal spectra and decompose an initial image into the union of nonintersecting sets. The second one is based on the transformation of an initial measure distribution under so called "direct multifractal transform" and calculation of information dimensions of measure supports. The examples of application of these techniques are given for various kinds of sea surface images.