Multisensory data fusion is one of the vibrant technologies, in which the data from several sources is acquired and processed together in order to get unified output. Data fusion systems are now widely used in various areas such as sensor networks, robotics, video and image processing, quality analysis of different products, intelligent system design etc.  Quality of fruits and vegetables consumed, contributes to health of human being. Generally, a team of trained sorters can do the manual quality analysis of fruits and vegetables by identifying its size, colour, smell, stiffness, skin texture etc. It takes much time and due to the subjectivity of individual team members, no uniform analysis would be possible by this way. The repetitive task of smelling may lead to the infection and/or irritation to the graders leading to degradation of the quality of grading. Use of Multisensory data fusion will be the good solution to get more correct and quick quality assessment of fruits and vegetables. In this technique, the acquired data of size, colour, smell, stiffness, and skin texture are fused using low level fusion to assess its quality. Data imperfection, spurious data, conflicting data, inconsistent data etc. are some of the issues that make data fusion a challenging task. There are a number of mathematical theories available to represent data imperfections, such as probability theory, fuzzy set theory, possibility theory, rough set theory, and Dempster–Shafer evidence theory (DSET). This paper presents the application of multisensory low level data fusion for quality assessment of fruits and vegetables using data of its colour and skin texture obtained from a machine vision camera and odour data obtained from an e-nose. The data collection system used in the study is described in detail along with the pre-processing and data fusion algorithms. Results obtained show usefulness of the technique for quality assessment of fruits and vegetables. It is reported in literature that studies that are focused on the fusion of digital images invariably showed improvements with respect to the results obtained using single data sources. The results obtain in present study indicate that the data fusion applied to machine vision and e-nose data complement well with each other and give positive correlation with the quality.

___________________________________________________________________________________

Keywords: Data Fusion, Sensors, E-nose, Machine Vision, Artificial Intelligence, Fuzzy Logic.