An Improved Multithreshold Segmentation Algorithm Based on Graph Cuts Applicable for Irregular Image

Table 1

Specific segmentation thresholds of different algorithms.

Image	level	Our algorithm	BA	MMSA	IBA	OTSU
LENA; (512×512)	2	121	118	120	121	119
	3	78,146	80,150	80,150	80,150	77,145
	4	58,108,160	60,109,160	60,109,160	60,109,160	56,106,159
	5	60,105,144,180	56,100,144,182	56,100,144,182	56,100,144,182	74,112,144,179
PEPPERS; (512×512)	2	128	128	130	129	126
	3	70,140	74,146	74,146	74,146	67,134
	4	61,115,164	61,112,164	72,135,193	61,112,164	61,117,165
	5	55,104,150,194	57,104,148,194	58,105,148,194	57,104,148,194	46,85,125,168
BIRD; (256×256)	2	105	108	106	109	104
	3	70,125	71,138	71,138	71,138	68,124
	4	67,128,177	70,129,177	70,129,177	70,130,177	65,116,159
	5	55,96,139,177	51,96,139,177	51,94,138,177	51,96,140,177	58,96,131,163
CAMERA; (256×256)	2	76	75	76	77	76
	3	71,150	128,193	128,193	128,193	69,143
	4	58,120,156	44,104,193	44,104,193	44,104,193	58,118,155
	5	40,95,146,195	44,97,146,197	44,97,146,197	44,97,146,197	41,94,139,169
GOLDHILL; (256×256)	2	115	113	116	115	114
	3	92,158	90,157	90,157	90,157	93,160
	4	80,126,178	79,132,178	79,132,178	79,132,178	82,125,178
	5	70,108,150,192	67,108,151,191	66,107,151,191	65,104,144,186	69,102,137,185