This paper proposes a novel data detection scheme based on the occurrence of blurred output images in image- sensor-based visible light communication systems. This system's receiver suffers from an inability to detect correct LED luminance values and demodulation of data because the captured image blurs owing to defocusing. The proposed method assumes that the light emitted by all LEDs that is diffused by the blur can be approximated by applying a Gaussian filter and expresses all LED luminance values as convolution equations based on this diffusion of the LED light. We estimate each LED light's condition by simultaneously solving these convolution equations and recover data according to the estimated result. To confirm the accuracy of the proposed method, we conduct computer simulations and an implementation experiment to evaluate the bit error rate performance of the proposed method.