铅浮选泡沫视频图像的物理校正及算法补偿

针对铅浮选视频采集图像光照不均、纹理丢失导致难以精确分割的问题,提出了一种用于优化视频图像的物理校正和算法补偿的方法。利用非成像光学理论构建均匀方形光斑自由曲面透镜,在此基础上,提出一种改进的方向波变换图像增强算法,再通过经典的谷底边缘检测算法对其进行分割。结果表明,该方法能有效提高浮选图像的信噪比和对比度,其分割精准度提高了25%。

Optical Correction and Compensation Algorithm for Flotation Image

As flotation images are difficult to segment due to the existence of uneven illumination, texture loss, a method of physical correction and compensation algorithms for image acquisition of lead flotation is developed. In this method, first, to construct the free-form surface lens based on the non-imaging optics theory is performed to improve the optical efficiency of the lens and the light source uniformity. Then, an improved directionlet transform image algorithm is proposed to increase the brightness in the small foam and to enhance the texture detail by analyzing the Retinex output characteristics of the shaded area and improving the gain function. Experimental results show that the method can improve flotation image SNR and contrast and the segmentation accuracy improved 25% by the classic bottom edge detection algorithm.