In this paper, an arrangement structure description method of multiclass objects based on polar coordinate feature matrix (PCFM) is proposed. The polar coordinate feature matrix consists of polar radius matrix (PRM) and Polar Angle Matrix (PAM), which describes the distance and angle information contained in the arrangement structure of multiclass objects. This method was applied in the application of vehicle fuse box detection and achieved obvious effects. Aiming at the vehicle fuse box detection, the method firstly uses a charge coupled device (CCD) industrial camera to acquire the images of the vehicle fuse boxes, employs histogram of oriented gradient (HOG) feature to characterize the ID codes of the vehicle fuse chips and combines SVMs to achieve the localization and recognition of the fuse chips in the images. According to the localization and recognition results, the PCFM is calculated to describe the arrangement structure of vehicle fuse chips in vehicle fuse box. Finally, the PCFM similarity is taken as the judgment criterion to realize the detection of the vehicle fuse box. The experiment proves that using PCFM similarity to detect the vehicle fuse box, the detection accuracy reaches 976%.