Search

Irregular text recognition of natural scene is a challenging task due to large span of character angles and morphological diversity of a word. Recent work first rectifies curved word region, and then employ sequence algorithm to complete the recognition task. However, this strategy largely depends on rectification quality of the text region, and cannot be applied to large difference between tilt angles of character. In this work, a novel anchor-free network structure of rotating character detection is proposed, which includes multiple sub-angle domain branch networks, and the corresponding branch network can be selected adaptively according to character tilt angle. Meanwhile, a curvature Adaptive Text linking method is proposed to connect the discrete strings detected on the two-dimensional plane into words according to people’s habits. We achieved state-of-the-art performance on two irregular texts (TotalText, CTW1500), outperforming state-of-the-art by 2.4% and 2.7%, respectively. The experimental results demonstrate the effectiveness of the proposed algorithm.

Background Cervicothoracic spinal tuberculosis is a rare disease. Due to its difficult and challenging surgical exposure, its surgical treatment approach remains inconclusive. Long-term follow-up studies to address this puzzling issue are rarely seen in the literature. The purpose of this study was to explore the selection of surgical treatment approaches for cervicothoracic spinal tuberculosis through a 10-year case review. Methods From January 2003 to January 2013, 45 patients suffering from cervicothoracic spinal tuberculosis were treated surgically. According to the relation between the tuberculosis lesion segments and the suprasternal notch on sagittal MRI, 19 patients were treated with a single-stage anterior debridement, fusion and instrumentation approach, and the other 26 patients were treated with a single-stage anterior debridement and fusion, posterior fusion and instrumentation approach. The clinical efficacy was evaluated using statistical analysis based on the Cobb angle of kyphosis, the Neck Disability Index (NDI) and the Japanese Orthopedic Association (JOA) scoring system. The neurofunctional recovery was assessed by the American Spinal Injury Association (ASIA) system. Results All patients were followed up for 6.6 years on average (range 3–13 years). No instrumentation loosening, migration or breakage was observed during the follow-up. The kyphosis angle and NDI and JOA scores were significantly changed from preoperative values of 34.7±6.8°, 39.6±4.6 and 10.7±2.8 to postoperative values of 10.2±2.4°, 11.4±3.6 and 17.6±2.4, respectively (p