Laser imaging entails scanning objects with a laser beam, generating an image based on the reflected beam. However, external noise and redundancy during scanning often distort the laser-reflected image characteristics, impeding recognition accuracy. To address this issue, we propose a laser imaging feature recognition method incorporating visual communication techniques. Specifically, the method first obtains laser imaging characteristics through a vision system calibration process. Subsequently, a convolutional neural network is utilized to denoise the laser imaging features. Finally, a support vector machine determines the classification boundary for the feature samples, and calculates the probability distribution of various samples from this boundary, enabling laser imaging feature recognition. Experimental results demonstrate that the proposed method achieves a recognition time of less than 100 seconds, and maintains a recognition accuracy of over 94% even in noisy conditions. This indicates the effectiveness of the proposed approach in improving the robustness and accuracy of laser imaging feature recognition.