To obtain the core body temperature of breeding pigs， a total of 108 female pigs from three breeds， Yorkshire， Landrace， and Yorkshire×Landrace hybrids were collected. A handheld infrared thermal imager was used to obtain infrared images of 11 body parts， including the eyes， ears， neck， shoulders， front back， hind back， rump， tail， genital area， hindquarters， and abdomen. Environmental information of the corresponding pig farm， including temperature， humidity， and wind speed， was obtained through temperature， humidity， and wind speed sensors. The data was divided into training and testing sets using a nested 5×4 cross-validation method. The preprocessed data was then used to build quantitative analysis models， including the least squares support vector regression （LSSVR）， support vector machine （SVM）， random forest （RF）， and ridge regression methods based on infrared image processing technology， as well as the local infrared imaging and environmental factors of breeding pigs. The LSSVR model was determined to be the best-performing model with a coefficient of determination （R2） of 0.639， and the root mean squared error （RMSE） and mean absolute error （MAE） were 0.133 and 0.110 ℃， respectively. To improve the model’s fitting effect， four possible influencing factors， including pig breed， pregnancy period， estrus， and sampling time （morning or afternoon）， were added. The results showed that except for pig breed， other factors increased the model’s performance by 4%， 8% and 10%， respectively. Finally， the R2 of the optimized model was 0.773 with an RMSE and MAE of 0.106 and 0.09 ℃， respectively. These results indicate that adding pregnancy period， estrus， and sampling time as factors can significantly improve the model’s fitting degree， making it more accurate and therefore useful as a factor for core body temperature inversion of breeding pigs.