The invention discloses a fruit
damage detection method and a device based on thermal characteristics of different tissues on the surface of a fruit. Temperature data of different regions of interest on a fruit sample are acquired through a
thermopile sensor; the temperature data are subjected to compression and
Fourier transform analysis sequentially and then curve peaks are extracted to serve as characteristic values; and the characteristic values are inputted to a
least squares linear classifier for classified training, and the to-be-detected fruit is detected. The device comprises a bracket, the
thermopile sensor, a
single chip microcomputer and a PC. A damage part is recognized by using the thermal characteristics of different tissues on the surface of the fruit, reliability is good, and accuracy is high.