During the production and processing of pineapple peeling line, inner thorns remain on the pineapple surface, which require manual secondary removal. To address this, an image processing method is adopted to extract features and spatially locate the inner thorns of pineapple, to determine their precise location for automatic removal of the thorns. Accordingly, a pineapple image acquisition system and a corresponding pineapple inner thorn detection algorithm are designed. The pineapple image is preprocessed to separate the inner thorn features from the background, eliminating the interfering features, and the contour of the inner thorn is used as the extracted feature. The area of the contour and the roundness of the contour are used as descriptors, and the center coordinates of the smallest outer rectangle of the pineapple inner thorn contour are used as the position of the pineapple inner thorn in the image, while the outer contour of the pineapple is used to convert the two-dimensional coordinates of the inner thorn into three-dimensional coordinates to precisely locate the pineapple inner thorn. The comparison experiments show that the accuracy of the pineapple inner thorn detection algorithm is significantly higher than that of the traditional speckle detection algorithm. In addition, the accuracy of fitting the center of the inner thorn is higher, with the maximum detection error of 0.63 mm and average detection error of 0.33 mm. The study shows that the average detection speed and accuracy can meet the needs of the pineapple inner thorn removal process, which provides a certain technical basis for the inner thorn removal in pineapple assembly line processing.