分析了傳統(tǒng)的基于馬爾可夫隨機(jī)場(chǎng)圖像分割算法收斂速度慢和固定加權(quán)等缺點(diǎn)，提出了一種基于簡(jiǎn)化馬爾可夫隨機(jī)場(chǎng)的紅外圖像快速分割算法。該算法首先對(duì)紅外圖像極大似然初始分割，并利用簡(jiǎn)化馬爾可夫隨機(jī)場(chǎng)對(duì)圖像進(jìn)行建模。在此基礎(chǔ)上采用自適應(yīng)的加權(quán)變化形式進(jìn)行迭代，不但加速了分割算法的收斂速度，而且使得分割效果都大為改善。在真實(shí)的飛機(jī)和艦艇紅外圖像上，該算法都取得了較好的分割效果。

　　Abstract:

　　This paper analyzes the shortcomings of the traditional MRF-based image segmentation algorithm with slow convergence and fixed-weighted， then a fast segmentation approach based on simple MRF for infrared images is proposed. The approach firstly performs random initial segmentation. By modeling the image with simple MRF and applying adaptive variable weighting forms while iterating， convergence is accelerated while improving the segmentation. Real aircraft and ships infrared images demonstrate the effectiveness of the approach.

?

?