基于RGB向量空间的彩色纹理图像检索

Colour Texture Image Retrieval Based on RGB Vector Space

彩色纹理图像分析是图像理解和计算机视觉研究领域的重点之一,彩色纹理图像分析具有很大的优越性。本文提出了一种新的彩色纹理图像描述子,首先计算RGB向量空间的梯度,经过标准化后把它映射到灰度级上,利用共生矩阵来描述纹理图像特征,并用于图像检索。实验结果表明,该纹理图像描述子能够有效结合颜色和纹理特征,具有良好的检索效果。

Colour texture image analysis is one of the key points of the image comprehendsion and the research sense of computer vision, the colour texture image analysis has very great superiority. This paper puts forward a new kind of colour texture descripter, computing the degreed of the RGB vector space gradient first, then is is standardized and projected to gray level, making use of the Co-occurrence Matrix to describe the image characteristic, and is used to image retrieval. Experimental results have shown that the proposed method has a good retrievall result, can combine the color and the texture characteristic effectively.