为了测量烤箱内烤鸡色泽品质并跟踪其变化过程，提出了一种炉膛烤鸡色泽区域取样方法以及量化检测色泽品质的模型。其中，色泽区域取样是通过基于水平集的自适应能量模型(ABEM)，以烤鸡图像的对比度信息和均值为基础，使用水平集方法对炉膛图像进行取样，实现烤鸡色泽区域图像的快速提取。为了验证该方法有效性，采用该方法和距离正则化水平集方法(DRLSE)对炉膛图像进行取样，试验结果表明，ABEM方法可在不损失取样效果的前提下快速获取炉膛烤鸡的前景图像。量化检测色泽品质的模型是通过基于色泽满意度的相对指标模型(CSIM)，利用HSV颜色空间中S分量的核心频域，计算烤鸡图像的色泽指标值，并转化为烤鸡相对色泽满意度指标，直观反映烤鸡色泽品质。试验结果表明将ABEM与CSIM相结合，对于监测炉膛烤鸡色泽品质指标是有效的，更为万能蒸烤箱的食品品质测控系统核心算法提供重要借鉴。

A new sampling method and model for quantitative measurement of color quality of oven-roasted chicken was investigated. An adaptive-balance energy model (ABEM), based on the level set was used for sampling. Based on the mean value and the contrast information from images of the roast chicken, the level set was employed for sampling images of the oven chamber, and rapid extraction of the color area in the roast chicken was achieved. In order to verify this, ABEM and distance-regularized level set evolution method (DRLSE) were adapted for sampling images of the oven chamber. The experimental results showed that the foreground image of the roast chicken in the oven could be obtained using ABEM without compromising the sampling. Using the model for quantitative determination of color quality, the color index of the roast chicken image was computed via a color satisfaction-based relative index model (CSIM) and the core frequency domain of S component in the hue-saturation-value (HSV) color space. Subsequently, the color index was converted to the relative color satisfaction index for roast chicken, directly reflecting the chicken color quality. The results indicated that a combination of ABEM and CSIM is effective in monitoring the color quality index for roast chicken, thus providing an important reference for the core algorithm in food quality control systems.

安徽省教育厅高等学校自然科学项目（KJ2012Z032）；国家火炬计划资助项目（2011GH031230）