Purpose. Two methods for image retrieval to facilitate understanding in biomimetic design were developed. The first increased precision of image returns and the second returned images by type. Method. Literature on image use towards comprehension and creativity are discussed to validate image use in the design process. Through this, it is shown that different types of images are better for different tasks. Text-based image retrieval is used rather than content-based image retrieval for reasons listed in the thesis. Search terms to increase precision are determined through natural language processing of glossary definitions. Search terms to return images of 'sequence', 'comparison', or 'drawn' type images are determined through examining textbooks, classifying images and applying words from their captions to Internet searches. Results . Noun-noun modifiers and head-nouns with sentence-structure depth greater than one from glossary definitions will typically increase precision. Words that return specific types of images are found.