基于图像阈值分割的改进蜂群算法

Improved Artificial Bee Colony Algorithm Based on Image Threshold Segmentation

为快速准确地将图像背景与目标进行有效分割,提出了一种基于图像阈值分割的量子改进蜂群算法(IABCQ:Improved Artificial Bee Colony Algorithm Based on Quantum)。该算法将量子比特概率幅的正弦分量引入到蜂群算法的编码中,通过调整相位角更新量子比特概率幅,使蜂群算法中引领蜂向当前最优蜜源的方向移动,避免算法搜索的盲目性;借鉴量子运算中非门操作将个体的正弦和余弦分量互换,使跟随蜂的蜜源进行互补更新;应用蜂群算法更新个数的限制,避免了局部优解和不动点引起的个体不更新问题。通过不同类型图像和算法之间的比较表明,该改进蜂群算法应用到图像阈值分割中的收敛时间减少了20%左右,同时也表现出良好的稳定性和抗噪声能力。

The image segmentation threshold is acquired to use gray-level images＇ information of pixels and histogram in order to segment between background and objects quickly and accurately. An IABCQ （Improved Artificial Bee Colony Algorithm Based on Quantum） in image threshold segmentation is proposed based on the quantum operation and basic ABC （Artificial Bee Colony） algorithm＇s mechanism. Firstly, this novel algorithm brings qubits probability amplitude＇s sinusoidal component into the encoding, then adjust the phase angle to update the qubits probability amplitude, which makes the employed bees move to the optimal nectar source to avoid the algorithm＇s search blindness. Secondly, the chromosomes＇ sine and cosine components are exchanged by quantum non-gate so that followed bees＇ nectar source can update complementarily. Thirdly, the limitation in the artificial bee colony is applied so as to avoid the local optimal solutions and fixed point. Finally, many types of images and algorithms＇ comparison can verify that convergent speed of this novel method in image threshold segmentation is reduced about 20% and this algorithm has good stability and anti-noise ability.