In order to solve the problem that the collimated target recognition algorithm of optical path docking cannot distinguish the adhesive state of double targets, a new method of collimated image dual optical target recognition based on Binary Large Object (BLOB) region feature analysis is proposed. There are two optical targets in the optical alignment image, that is, the simulated optical target and the main laser target. In the initial beam control stage, the positions of the two optical targets are random and uncertain, and there is a possibility of the two optical targets sticking together, which causes great difficulties in beam control. Therefore, optical path alignment needs to solve the image recognition problem in two cases: 1) In the initial beam control stage, when the main laser beam and the analog beam are just introduced, the adhesion recognition algorithm needs to be used to identify the adhesion state of the two optical targets. If the two optical targets are in the adhesion state, the two targets need to be completely separated by adjusting the 2D frame BM6XY motor; 2) In the case of two optical targets completely separated, it is necessary to distinguish between the analog light target and main laser target in the two optical targets. Firstly, the binary image is processed by digital morphology to calculate the area, center C_xy, axis length len_xy and region Regin_xy, of each BLOB region in the whole image. Secondly, the number of valid BLOB regions vblobcount is counted, and the distance between the two maximum connected domains dir is calculated. When vblobcount>1 and dir>100, the collimation image is the completely separated double target image, otherwise it is the adhered image. Then, for the completely separated dual-target image, the number of BLOBs located in the two largest BLOB regions with the center of each BLOB region is counted. The small number of candidate BLOB regions is the main laser target, and the large number of candidate BLOB regions is the analog light target. Finally, for the adhered image, when dir<100, the two optical targets are in a close adhered state. The axis length coefficient len_xs is calculated. When len_xs≥1.2, the two targets are in a very close adhesion state, otherwise it is a single main excitation target or simulated light target. Moreover, the repetition accuracy analysis of the adhesive optical target recognition method based on edge feature analysis showed that the correlation coefficients of the adhesive image in the left, right, up and down directions are 0.967 1, 0.990 0, 0.999 7 and 0.922 4, respectively. The mean value of the correlation coefficients is 0.969 8 and the mean square error was 0.034 4. Among them, the correlation coefficients in the left, right and upper directions are greater than 0.95, which are marked by red, green and yellow curves respectively. The mean value of the correlation coefficients in the left, right and upper directions is 0.985 6, and the mean square error is 0.016 7. Since the third type of adhesive image is mainly the adhesive image formed after the laser target and the simulated optical target are migrated in the vertical direction, the overlap area is mainly located in the upper half of the adhesive image. The experimental results show that the proposed dual optical target recognition algorithm can not only realize the recognition of completely separated analog light targets and main laser targets with the error less than 3 pixels and the processing time less than 1 s, but also realize the identification of optical targets in the state of adhesion and a single independent optical target. It meets the requirements of self-adaptability, accuracy and efficiency of the optical alignment image recognition algorithm and the repetition accuracy of the recognition of different adhesive images, and can solve the problem of the optical alignment target recognition algorithm to accurately identify the adhesion state of two targets