自然界鸟类图像特征准确识别仿真研究

摘要

It can enhance image retrieval efficiency to accurately identify image features of nature birds.Image gray level correction decides the accuracy of image feature recognition.It needs to obtain gray level histogram of image pixel in time domain after decomposing during correcting image gray level.The gray level histogram is used to correct the gray level.Traditional method disposes image by homomorphic filtering method,it accomplishes gray level correction via compressing gray level range and enhancing image contrast ratio quality.However,it cannot build gray level mapping relation and reduces updated accuracy.In this paper,we propose an image gray level correcting method based on modified Wavelet Transform (WT) to identify image features for nature birds.Optics imagery system theory is fused to demarcate linear array CCD system,like optics shot and mechanical structure.Then the coefficients are corrected and the image is resolved according to orthogonal WT,and the histogram of image pixel in time domain is obtained.Finally,the rank mid-value is got,and the gray level histogram is used to correct image gray level.Simulation results show that the above method has strong robustness to gray level changing,and can improve the recognition accuracy of image features for nature birds.%对自然界鸟类图像特征进行准确识别,可提高图像的检索效率.图像灰度校正决定着鸟类图像特征识别的准确度,进行图像灰度校正时,需要获取分解后图像像素的时域灰度直方图,利用直方图中的灰度映射进行灰度校正,而传统方法利用同态滤波方法对图像进行处理,利用压缩灰度范围和增强对比度增强图像的质量完成灰度校正,但不能建立灰度的映射关系,降低了校正的精度.提出一种改进小波变换的图像灰度校正方法,对自然界鸟类图像特征进行识别.上述方法融合于光学成像系统原理对光学镜头、机械结构在内的线阵CCD系统进行标定,得到校正系数,依据正交小波变换分解图像,并获取分解后图像像素的时域直方图,并对排序得到中值,利用其中值直方图灰度映射完成对图像灰度校正.仿真结果表明,所提算法对灰度变化的校正具有较高的稳健性,可以提高自然界鸟类图像特征识别的准确率.