The accessibility of website images is influenced by the availability and accuracy of descriptive text and its compatibility with image complexity and purpose. Image accessibility evaluation cannot be fully affected through applying one method, and can be enhanced by inclusion of processes that consider the quality of image descriptive text. The evaluation of descriptive text quality may initially involve human evaluation, then use an automated evaluation tool to provide a counterpoint. In this paper, an analysis is presented of a dataset of 120 complex and informative images found on universities’ web-based systems. This is supplemented with a detailed analysis of HTML image attributes and elements. Human and automated analysis of content are combined and the integration of information to inform the evaluation outcome. Our analysis illustrates a lack of accurate usage of HTML image attributes and elements, such as alt and longdesc. The findings provide insight for the improvement of image accessibility by applying multi-evaluation methods and auto generator descriptive text. This paper will be of interest to web accessibility developers and researchers.