倾斜图像矫正算法的研究

Research on Tilted Image Correction Algorithm

为提高倾斜图像矫正效果、降低计算复杂度,本文深入分析了TILT (Transform invariant low-rank textures)算法并对其进行优化改进,包括凸优化算法、拉格朗日乘子法、多分辨策略优化方法等。为了验证算法有效性,本文对车牌、建筑、文字以及人脸等不同类型的倾斜图像进行矫正实验。仿真结果表明,TILT及其改进算法对多种角度甚至是较大角度倾斜图像的矫正都可获得较为理想结果,且改进算法可有效降低计算复杂度,因而该算法在图像的后续处理、识别等领域具有广泛的应用前景。

In order to improve the tilt image correction effect and reduce the computational complexity, this paper analyzes and optimizes the TILL (Transform invariant low-rank textures) algorithm, including the convex optimization algorithm, the Lagrange multiplier algorithm, the multi-resolution strategy optimization algorithm, etc. In order to verify the validity of the algorithm, this paper corrects different types of tilt images, such as license plate, building, text and face. The simulation results show that TILT and its improved algorithm can obtain more ideal results for the correction of tilt images from various angles even larger angles, and the improved algorithm can effectively reduce the computational complexity, so the algorithm has a wide range of applications in the fields of image follow-up processing and recognition.