Understanding of the sensory representation in neuronal population is a major problem in the sensory neuroscience. In the primary visual cortex (V1), response properties of individual neurons are well-known, whereas information in a population of neurons is less examined. Here, we studied visual representation of neuronal population in the mouse V1. Using two photon functional calcium (Ca) imaging, neuronal activities were recorded from the V1 layer 4 neurons in anaesthetized mice. To estimate the information in the activities, we reconstructed natural image stimuli from the Ca signals and evaluated how exactly the visual images were predicted by simultaneously-recorded, several hundred neurons' activities. To reconstruct visual images, a liner filter which directly relates the neuronal activity to the visual image was computed by a least square method with regularization. The reconstruction filter was estimated with 90 % visual images and corresponding activity data, and its performance was tested with the remaining 10 % of new images and activity data (test data). We found that visual images were roughly reconstructed by a single trial activity; low pass filtered-like, blurred images appeared in the reconstructed images. Correlation coefficients (r) between stimulus and reconstructed images were 0.31 (0.22-0.40) for single trial based image and 0.52 (0.43-0.59) for trial-averaged image (n=5 populations from four mice). Then we examined the effect of correlated activity between neurons on the reconstruction. The reconstruction filter was estimated with trialsequence-shuffled data. The reconstruction was improved by this data manipulation (r = 0.32 for original test data and 0.36 for trial-shuffled test data in single trial based reconstruction). These results suggest that natural images can be reconstructed by activities of simultaneously recorded several hundred neurons and the correlated activities may reduce the information among neurons in V1. We are examining the reconstruction based on the receptive field of individual neurons, and present how the receptive field information affects the image reconstruction.