可重叠矩形多值图像表示及其上的几何矩生成

Overlapping Rectangle Multi-valued Image Representation and its Application in Geometric Moments Generation

为了支持快速的多值图像运算,提出了一种无损多值图像表示方法,称为可重叠矩形多值图像表示(Overlapping Rectangle Multi-valued I mage Representation,ORMIR)。ORMIR采用递归方式将一幅多值图像分割为具有不同基础颜色的可重叠矩形,并使用孩子兄弟树来组织这些矩形,通过弱化二值图像块表示中同一矩形所覆盖的所有像素必须具有相同颜色的约束,ORMIR能够使用较少的矩形无损地表示一幅多值图像,因而基于ORMIR的多值图像运算能够被快速实现。基于ORMIR,提出了一个多值图像几何矩生成算法,该算法首先生成多个仅包含一个矩形区域的二值图像的几何矩,然后将这些几何矩加权求和得到原始多值图像的几何矩。试验结果表明,基于ORMIR的几何矩生成算法能够以每秒50帧以上的速度计算8比特位深的512×512的灰度图像直到3+3阶的几何矩,从而满足实时应用的需要。

A lossless multi-valued image representation, referred as to Overlapping Rectangle Multi-valued Image Representation （ORMIR） was presented in this paper for the fast generation of geometric moments of multi-valued images. ORMIR divides a multi-valued image into some overlapping rectangles with some basic colors and settles these rectangles in a child-sibling tree. By impairing the constraint in Image Block Representation where all the pixels covered with a distinct rectangle must have the same color, ORMIR has reduced significantly the number of the rectangles required to represent a multi-valued image, hence is able to speed up image operations such as geometric moment generation. Based on ORMIR, a fast algorithm to generate geometric moments of multi-valued images was presented, which firstly generates all the geometric moments of a corresponding set of binary images with only one rectangular region and then sums up these moments by some weight in order to obtain the geometric moments of the original multi-valued image. The experimental results show that the algorithm is capable of generating the geometric moments up to order 3＋3 of a 512× 512 8 bits gray image beyond a speed of 50 frames per second and therefore, it is competent to geometric moments generation in real-time applications.