The wavelet coefficients of an image will obtain higher performance of energy concentration after its high frequency components are decreased by low pass filters. This paper makes a study of the near lossless compression scheme based on this principle. In order to make smoothing process more effective to improve compression rate of the above near lossless compression scheme

this paper presents a novel smoothing algorithm

which can achieve better smoothing performance by adaptively select appropriate filter direction for every pixel according to its horizontal and vertical gradient intensity in the process of scan. Another remarkable advantage of this algorithm is that it can smooth the image according to given maximum amplitude error

then makes the accuracy of near lossless compression adjustable. In addition

this algorithm is easy to implement for its low computation complexity. The experimental results with several standard test images demonstrate the effectiveness and adaptability of the presented algorithm in this paper.