In order to find the method of auto screening body temperature in pig’s cultivation, the thermal infrared imager was used to acquire the visible light image and infrared thermograph of pig at the same time from side-view. A series of registration and fusion processes were taken for the two kinds of images. A method which was based on improved active shape model (ASM) was proposed to detect the ear area of pig. Firstly, considering the characteristic of the pig ear profile, the marking rules of landmark points were restricted. Furthermore, the non subsampled contour transform (NSCT) coefficients of fusion image were substituted for the normalized first derivatives of profiles to build the gray level appearance model of each landmark. Finally, the process of average shape initialization was adapted based on matching skeleton endpoints. 50 images were tested by using the proposal method. The test result of overlap ratio greater than 0.8 accounted for 84%. The proposal method is significantly valuable on extracting the ear root area from recognized ear contour and developing the system of auto screening body temperature based on infrared thermograph.