According to the problem of the low matching accuracy rate and the low speed in the matching technology for image search, we choose the image of the strawberry diseases as the object of the research, to extract the feature of strawberry diseases image and then do cluster analysis using SPSS, and choose the feature combination according to the clustering effect. Do search matching experiments on the feature combination by KD-Tree matching algorithm respectively, and determine the characteristics combination which search matching accuracy is higher by comparing the results. This feature combination of strawberry powdery mildew matching accuracy reached 83.3%, the deformity of strawberry matching accuracy reached 60.0%, the aim is to provide the research basis for rapid disease diagnosis on strawberry.