Invertebrate pests are difficult to control and the losses caused by them is huge. In six major Australian grain crops, the estimated annual loss from invertebrate pests is $359.8 million [1, 2]. Pests include molluscs, insects and nematodes all at different stages of their life cycle. Traditionally, light traps and hand nets were used to sample the pests that are capable of flight, and human experts and recent machine vision systems were used to classify the samples. This approach is labour intensive and only captures insect pests at one late stage of their life cycle, which may be too late for integrated pest management (IPM). IPM uses the combination of all possible pest control methods to reduce the amount of insecticide must be used, providing advantages to both the environment and the consumers. IPM would be much easier if there were a technological capability to detect pests on the crop in all stages of the lifecycle of both crops and pests. Many pests have defence mechanisms that involve camouflage of colour and shape. Predators of pests, which includes insects, arachnids and birds have evolved techniques for detecting prey. Besides that, some insects, such as butterflies, are very good at finding healthy leaves. Inspired by the vision of predators of invertebrate pests and leaffeeding insects, we have developed a multispectral 3D vision system that can detect common invertebrate pests on green leaves. The main contribution of this study is that we proposed a high-dimensional colour space named hyper-huesaturation- intensity (HHSI), which is less affected by unstable illumination and could enlarge inter-class distance for material classification.