To address the issue of low positioning accuracy caused by incomplete image capture of light emitting diodes(LEDs), a novel indoor visible light positioning algorithm based on visual error compensation is proposed. This algorithm consists of three components: a pixel coordinate fitting and restoration algorithm, a cyclic coordinate optimization algorithm, and a dual-lamp positioning algorithm that compensates for height differences. These methods are designed to handle rectangular luminaire images captured under various occlusion conditions, reduce the impact of image noise on positioning results, and improve positioning accuracy in scenarios with tilted ceilings. The experimental results show that compared to positioning accuracy under unobstructed conditions, the positioning error using the pixel coordinate fitting and restoration algorithm is less than 1.5 cm. The cyclic coordinate optimization algorithm can compensate for the effects of image noise on positioning, reducing the positioning error by 60%. Additionally, the dual-lamp positioning algorithm that compensates for height differences reduces the average positioning error by 65%.