JPEG图像压缩对测量类图像的影响

Influence of JPEG compression on images for measuring

在JPEG图像压缩的量化过程中,丢弃了部分高频信息,使图像的清晰度下降。通过对频谱的对照分析和对实际图像进行实验后表明,JPEG图像压缩对测量和处理结果的影响与原图像的特点、取样方式、滤波过程和压缩算法有关。经摄像机和图像卡获得的图像,其压缩比可达到10;经扫描仪获得的图像,其压缩比可达到6;经数码相机获得的图像,其压缩比可达到6～8;对特定的医学图像、晶相图像等来说,利用图像的特征构造最优匹配的JPEG压缩量化表,其压缩比可达到10。采用合适的压缩比和改进的压缩方法,图像的细节损失可以控制在较小的范围内,基本上不影响检测类图像的边界提取、分割、尺寸测量时的处理结果。

During the quantization of JPEG compression, a part of high frequency information is lost. As result, the image's definition decreases. Through frequency spectrum analysis and factual experiment, it is found that JPEG compression's influence on measurement and consequence is related to original image's character, sample method, filter processing and compression algorithm. For images derived from vidicon and capture card, compression ratio could be up to 10. For images derived from scanner, it could be up to 6. For samples from digital camera, compression ratio could be up to 6-10. With appropriate compression ratio and improved compression method, the loss of the image could be within a small range. On the whole, the loss does not affect edge detection, division, measurement and so on. The conclusion provides evidence for high-resolution image compression method's application in dimension and position detection.