TKIS-I helmet-mounted colorful imaging sonar is mounted on the helmet of divers to serve as their underwater eyes. Currently, there are more than two dozens of it serving the navy of China. However, under the complex underwater environment, divers usually take great risks while performing underwater operations. The work of this paper aims to achieve automatic underwater target detection of the imaging sonar in order for divers not to dive into water. The paper combined Histogram of Oriented Gradient (HOG) in computer vision for feature extraction and support vector machine SVM) for classification to achieve quick underwater target detection. The results showed high detection rate. All these work are the foundation for automatic underwater target detection and recognition in the future.