A machine vision system was used to identify weeds commonly found in Kansas wheat fields, including Russian thistle, redroot pigweed, Palmer amaranth, wild buckwheat, and kochia. Three different approaches, color analysis, shape analysis, and texture analysis, were used in the study. For the color analysis approach, ratios of pixel gray levels in images taken using four selected color filters were useful in classifying pixels into five different categorieswheat leaf, weed leaf, weed stem, soil, and sand. A red/green filter pair was found effective in identifying reddish stems of redroot pigweed, Russian thistle, and kochia. Five shape parameters, eccentricity, compactness, and three invariant moments, were used in leaf shape studies and were found effective in distinguishing broadleaf weed species such as redroot pigweed, wild buckwheat, and kochia from wheat. For the texture analysis approach, Fourier spectra of selected windows within leaf areas of wheat and weed species were analyzed. An index of fineness was defined using the spectra. Leaves with fine textures such as kochia can be distinguished from others using this index. Curves of normalized radial spectral energy were derived from the spectra. Leaves with distinct directionality features such as wheat and some broadleaf weed species can be distinguished using parameters defined using these curves. This study is the first step of a project with an overall goal of developing techniques of selective herbicide application based on weed detection.